|Publication number||US20090012984 A1|
|Application number||US 11/968,433|
|Publication date||8 Jan 2009|
|Filing date||2 Jan 2008|
|Priority date||2 Jul 2007|
|Also published as||US8825673, US8938461, US20100198864, US20100287466, US20150098660, WO2009004624A2, WO2009004624A3, WO2009004624A4|
|Publication number||11968433, 968433, US 2009/0012984 A1, US 2009/012984 A1, US 20090012984 A1, US 20090012984A1, US 2009012984 A1, US 2009012984A1, US-A1-20090012984, US-A1-2009012984, US2009/0012984A1, US2009/012984A1, US20090012984 A1, US20090012984A1, US2009012984 A1, US2009012984A1|
|Inventors||Yiftach Ravid, Amir Milo|
|Original Assignee||Equivio Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Referenced by (25), Classifications (10), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application claims the benefit of U.S. Provisional Application No. 60/974,974 filed on Sep. 25, 2007, and U.S. Provisionial Application No. 60/947,606 filed on Jul. 27 2007, both of which are hereby incorporated by reference herein.
The present invention relates to the field of organizing large numbers of documents.
In litigation proceedings, as well as for other functions, often massive amounts of documents must be reviewed. Certain organizational methods for arranging documents exist in the art. Emails are a particular type of document that are useful to review in structures, to help make sense of the proceedings and reduce the number of documents that need to be read.
The need to detect near duplicate documents arises in many applications. Typically this may occur in litigation proceedings. In litigation, often one of the parties initiates discovery proceedings which force the rival party to reveal all the documents at his disposal that pertain to the legal dispute. In order to meet the provisions of the discovery procedure, the disclosing party hands piles of documents, sometimes in order to duly meet the full disclosure stipulations, or in certain other cases, as a tactical measure to flood the other party with huge amounts of documents, thereby incurring the receiving party considerable legal expenses in the tedious task of determining which documents are relevant to the dispute under consideration. In many cases, out of the repertoire of disclosed documents, many are similar to each other. A preliminary knowledge which will group and/or flag documents that are similar to one another would streamline the screening process, since for example, if a certain document is classified as irrelevant, then probably all the documents that are similar thereto, are also deemed irrelevant. There are numerous other applications for determining near duplicate documents, sometimes from among a very large archive of documents (possibly of the order of millions of documents or more).
A common type of document that is examined in litigation procedures is emails. If collected from user accounts of various users in a company, there is likely to be a degree of duplicity between users. Duplicity may occur because the same email is sent to a number of recipients at once, or for other reasons. Also, many times, emails are near duplicates of one another.
U.S. Pat. No. 7,035,876 to Kawai et al provides a system and method for evaluating a structured message store for message redundancy. A header and a message body are extracted from each of a plurality of messages maintained in a structured message store. A substantially unique hash code is calculated over at least part of the header and over the message body of each message. The messages are grouped by the hash codes. One such message is identified as a unique message within each group. In a further embodiment, the messages are grouped by conversation thread. The message body for each message within each conversation thread group is compared. At least one such message within each conversation thread group is identified as a unique message. The invention requires that all emails in a set have the same subject line. Additionally, all emails must have the same attachment to be considered part of the same set.
U.S. Pat. No. 6,119,124: Method for clustering closely resembling data objects. A computer-implemented method determines the resemblance of data objects such as Web pages. Each data object is partitioned into a sequence of tokens. The tokens are grouped into overlapping sets of the tokens to form shingles. Each shingle is represented by a unique identification element encoded as a fingerprint. A minimum element from each of the images of the set of fingerprints associated with a document under each of a plurality of pseudo random permutations of the set of all fingerprints, are selected to generate a sketch of each data object. The sketches characterize the resemblance of the data objects. The sketches can be further partitioned into a plurality of groups. Each group is fingerprinted to form a feature. Data objects that share more than a certain numbers of features are estimated to be nearly identical.
U.S. Pat. No. 6,189,002: Process and system for retrieval of documents using context-relevant semantic profiles. A process and system for database storage and retrieval are described along with methods for obtaining semantic profiles from a training text corpus, i.e., text of known relevance, a method for using the training to guide context-relevant document retrieval, and a method for limiting the range of documents that need to be searched after a query. A neural network is used to extract semantic profiles from text corpus. A new set of documents, such as World Wide Web pages obtained from the Internet, is then submitted for processing to the same neural network, which computes a semantic profile representation for these pages using the semantic relations learned from profiling the training documents. These semantic profiles are then organized into clusters in order to minimize the time required to answer a query. When a user queries the database, i.e., the set of documents, his or her query is similarly transformed into a semantic profile and compared with the semantic profiles of each cluster of documents. The query profile is then compared with each of the documents in that cluster. Documents with the closest weighted match to the query are returned as search results.
U.S. Pat. No. 6,230,155: Method for determining the resemblance of documents. Disclosed is a method for facilitating the comparison of two computerized documents. The method includes loading a first document into a random access memory (RAM), loading a second document into the RAM, reducing the first document into a first sequence of tokens, reducing the second document into a second sequence of tokens, converting the first set of tokens to a first (multi)set of shingles, converting the second set of tokens to a second (multi)set of shingles, determining a first sketch of the first (multi)set of shingles, determining a second sketch of the second (multi)set of shingles, and comparing the first sketch and the second sketch. The sketches have a fixed size, independent of the size of the documents. The resemblance of two documents is provided, using a sketch of each document. The sketches may be computed fairly fast and given two sketches, the resemblance of the corresponding documents can be computed in linear time in the size of the sketches.
U.S. Pat. No. 6,240,409: Method and apparatus for detecting and summarizing document similarity within large document sets. A method and apparatus are disclosed for comparing an input or query file to a set of files to detect similarities and formatting the output comparison data are described. An input query file that can be segmented into multiple query file substrings is received. A query file substring is selected and used to search a storage area containing multiple ordered file substrings that were taken from previously analyzed files. If the selected query file substring matches any of the multiple ordered file substrings, match data relating to the match between the selected query file substring and the matching ordered file substring is stored in a temporary file. The matching ordered file substring and another ordered file substring are joined if the matching ordered file substring and the second ordered file substring are in a particular sequence and if the selected query file substring and a second query file substring are in the same particular sequence. If the matching ordered file substring and the second query file substring match, a coalesced matching ordered substring and a coalesced query file substring are formed that can be used to format output comparison data.
U.S. Pat. No. 6,349,296: Method for clustering closely resembling data objects. A computer-implemented method determines the resemblance of data objects such as Web pages. Each data object is partitioned into a sequence of tokens. The tokens are grouped into overlapping sets of the tokens to form shingles. Each shingle is represented by a unique identification element encoded as a fingerprint. A minimum element from each of the images of the set of fingerprints associated with a document under each of a plurality of pseudo random permutations of the set of all fingerprints, are selected to generate a sketch of each data object. The sketches characterize the resemblance of the data objects. The sketches can be further partitioned into a plurality of groups. Each group is fingerprinted to form a feature. Data objects that share more than a certain numbers of features are estimated to be nearly identical.
U.S. Pat. No. 6,658,423: Detecting duplicate and near-duplicate files. Disclosed is an improved duplicate and near-duplicate detection. Techniques may assign a number of fingerprints to a given document by (i) extracting parts from the document, (ii) assigning the extracted parts to one or more of a predetermined number of lists, and (iii) generating a fingerprint from each of the populated lists. Two documents may be considered to be near-duplicates if any one of their respective fingerprints match.
U.S. Pat. No. 6,654,739: Lightweight document clustering is a procedure for clustering documents that operates in high dimensions, processes tens of thousands of documents and groups them into several thousand clusters or, by varying a single parameter, into a few dozen clusters. The procedure is specified in two parts: computing a similarity score representing the k most similar documents (typically the top ten) for each document in the collection, and grouping the documents into clusters using the similar scores.
U.S. Pat. No. 6,751,628: Process and system for sparse vector and matrix representation of document indexing and retrieval. Disclosed is a new data structure and algorithms which offer at least equal performance in common sparse matrix tasks, and improved performance in many. This is applied to a word-document index to produce fast build and query times for document retrieval.
U.S. Pat. No. 7,139,756: System and method for detecting duplicate and similar documents. A system and a method are described for rapidly determining document similarity among a set of documents, such as a set of documents obtained from an information retrieval (IR) system. A ranked list of the most important terms in each document is obtained using a phrase recognizer system. The list is stored in a database and is used to compute document similarity with a simple database query. If the number of terms found to not be contained in both documents is less than some predetermined threshold compared to the total number of terms in the document, these documents are determined to be very similar. It is shown that these techniques may be employed to accurately recognize that documents, that have been revised to contain parts of other documents, are still closely related to the original document. These teachings further provide for the computation of a document signature that can then be used to make a rapid comparison between documents that are likely to be identical.
Abdur Chowdhury Duplicate Data Detection The algorithm is based on IDF of the tokens. The algorithm steps are: 1. Get document. 2. Parse document into a token steam, removing, format tags. 3. Using term thresholds (idf), retain only significant tokens. 4. Insert relevant tokens into Unicode ascending ordered tree of unique tokens. 5. Loop through token tree and add each unique token to the SHA1 (1995) digest. Upon completion of token tree loop, a (doc_id, SHA1 Digest) tuple is defined. 6. The tuple (doe_id, SHA1 Digest) is inserted into the storage data structure based on SHA1 Digest key. 7. If there is a collision of digest values, then the documents are similar. Conrad et. A1: In a series of a few papers, they describe a method that is based on the IDF measure of tokens, and the size of the documents. They are also provided a method of selecting the corpus to evaluate the IDF of a token.
There is thus a need in the art to provide for a new system and method for determining near duplicate objects. There is still further need in the art to provide for a new system and method for determining near duplicate documents.
The present invention relates to the organization and display of data, particularly when the source of the data is a huge number of documents.
According to an aspect of the invention, there is provided a computer product including a data structure for organizing of a plurality of documents, and capable of being utilized by a processor for manipulating data of the data structure and capable of displaying selected data on a display unit. The data structure comprises:
a) a plurality of directionally interlinked nodes, each node being associated with at least one document having at least a header and body text; and wherein all documents associated with a given node having substantially identical normalized body text, and wherein all documents having substantially identical normalized body text being associated with the same node, and wherein at least one node being associated with more than one document;
b) for any first node of the nodes that is a descendent of a second node of the nodes, the normalized body text of each document associated with the first node is substantially inclusive of the normalized body text of each document that is associated with the second node.
According to a further embodiment of the invention, all documents associated with a given node further have substantially identical normalized subject parameter in the header.
According to a further embodiment of the invention, there is provided that the body text of each document associated with said first node is substantially inclusive of the body text of each document that is associated with said second node, irrespective of whether a normalized subject parameter from a header of a document associated with said first node and a normalized subject parameter from a header of a document associated with said second nodes are identical.
According to another aspect of the invention, there is provided a method for organizing documents into nodes, in which a node represents a group of substantially equivalent documents. The method comprises:
(i) providing a plurality of original documents, each comprising a header and a body, and wherein the header comprises at least one parameter and wherein the body comprises text,
(ii) selecting a document from among the documents and associating the document with a node, comparing at least a portion of the body text of the document to at least a portion of the body texts of other documents from amongst the plurality of documents, and in the case of a match, merging the node associated with the document with a node associated with the matching document,
(iii) searching the body of the document to locate a first instance of header-type text, wherein the header-type text contains at least one header parameter;
(iv) constructing a presumed document comprising a header and a body, wherein the header of the presumed document comprises one or more parameters from the header-type text located within the body of the original document, and wherein the body of the presumed document substantially comprises the text located after the header-type text in the body of the original document, and associating the presumed document with a node;
(v) comparing at least a portion of the body text of the presumed document to at least a portion of the body texts of at least one other documents from among the plurality of documents and in the case of a match, merging a node associated with the presumed document with a node associated with the matching document.
(vi) if the comparison of (v) does not find a match, processing repeatedly the remainder of the body of the document for successive instances of header-type text, as stipulated in stages (iii)-(v), and for each instance, constructing a presumed document, comparing for any matching documents to the presumed document, and if found, merging the nodes associated with the matching documents, until no new presumed documents are found.
According to another aspect of the invention, a method is provided for reducing duplicate document display of a large number of documents, in which the method involves:
a) comparing a fingerprint of a document with previously stored document fingerprints, wherein a fingerprint is formed for each of at least a portion of the normalized body text and a normalized subject parameter of a document, wherein the comparison for detecting and indicating duplicating documents;
b) searching the document for instances of header-type text, searching in text order through the normalized body text of the document, and if header-type text is found in the search,
In an embodiment of the invention, the method described above also involves providing a plurality of nodes, and associating each document having a unique fingerprint with a unique node, and associating each document detected as duplicating to a prior document with the node associated with the prior document.
In an embodiment of the invention, the method described above also involves linking nodes to provide that a node associated with a first presumed document becomes the parent of the node associated with the document from which the first presumed document is derived, and to provide that the node associated with each sequentially derived presumed document derived from the same document becomes a parent of the node associated with the previously derived presumed document.
In an embodiment of the invention, the method described above also involves displaying the nodes in a computer format, and affiliating each node with the body text and subject parameter of the document associated with the node.
According to another aspect of the invention, there is provided a computer product including a data structure for organizing of a plurality of documents, and capable of being utilized by a processor for manipulating data of the data structure and capable of displaying selected data on a display unit. The data structure comprises one or more trees, wherein a tree comprises at least a trunk and at least one node, each node being associated with a document having at least a header and body text, and wherein a trunk being associated with zero or more documents having at least a header and a body text and wherein all documents whose body text includes the same included document are associated with the same tree, and wherein a unique inclusive document, as well as documents that duplicate to said unique inclusive document, are associated with one of one or more unique nodes of said tree, and wherein an included document, as well as documents that duplicate to said included document, are associated with the trunk of the tree.
In order to understand the invention and to see how it may be carried out in practice, an example embodiment will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:
It should be noted that the invention is described for convenience, with reference to documents. The term documents is understood to include files including text or representing text, such as Microsoft Word documents, Excel documents, mail documents, etc. References to documents embrace also derivative thereof, such as known per se canonic representation of a document. In accordance with certain embodiments, documents include at least text and/or numbers. In one embodiment, the documents are Microsoft OfficeŽ documents, such as e-mails in selected format. The format may be, for example, Microsoft Outlook, Lotus Notes, etc. The term documents is not confined to obvious documents, but applies also to other types of data objects, such as documents within a ZIP file, e-mails in MS Outlook PST file format, attachments, etc.
In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be understood by those skilled in the art, that the present invention may be practiced without these specific details. In other instances, well-known methods, procedures, components and circuits have not been described in detail so as not to obscure the present invention.
Unless specifically stated otherwise, as apparent from the following discussions, it is appreciated that throughout the specification discussions, utilizing terms such as, “processing”, “comparing”, “linking”, “connecting”, “representing”, “detecting”, “searching”, “deriving”, “calculating”, “storing”, “inserting”, “determining”, “treating”, “repeating”, “identifying”, “labeling”, “indexing”, “running an algorithm”, “to return, if positive . . . if negative . . . ”, “creating”, “showing”, “displaying”, “suppressing”, “setting levels”, “stringing nodes”, “organizing”, “associating”, “affiliating”, or the like, refer to the action and/or processes of a computer or computing system, or processor or similar electronic computing device, that manipulate and/or transform data represented as physical, such as electronic, quantities within the computing system's registers and/or memories into other data similarly represented as physical quantities within the computing system's memories, registers or other such information storage, transmission or display devices.
Embodiments of the present invention may use terms such as processor, computer, apparatus, system, sub-system, module, unit and device (in single or plural form) for performing the operations herein. Devices may be specially constructed for the desired purposes, or the invention may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, magnetic-optical disks, read-only memories (ROMs), random access memories (RAMS) electrically programmable read-only memories (EPROMs), electrically erasable and programmable read only memories (EEPROMs), magnetic or optical cards, or any other type of media suitable for storing electronic instructions, and capable of being coupled to a computer system bus.
The processes/devices (or counterpart terms specified above) and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct a more specialized apparatus to perform the desired method. The desired structure for a variety of these systems will appear from the description below. In addition, the present invention is not limited to any particular programming language or type. It will be appreciated that a variety of programming languages may be used to implement the teachings of the inventions as described herein.
The following description seeks to explain by way of example what inclusiveness is, in relation to emails. With reference now to
As has been noted, body text 23 is inclusive of body text 22, which is inclusive of body text 21. Thus, if one wished to read the minimum number of texts, one could suffice by reading only body text 23. A meaning of the term ‘inclusive document’ in the present context is directed to one document from amongst a group of documents, in which the inclusive one includes the text (or the normalized text) of all the others. In the set of documents of
It will be noted in certain embodiments, when determining inclusiveness, that a first document is only considered inclusive of a second document if the first document contains within its body text the entire body text of the second document. However, a document can still be considered inclusive if certain minor differences are present, such as for example, normalization having been applied, or legal disclaimers at the end of an email having been stripped, or only a certain number of lines or characters of the body text are considered as significant.
Having described what inclusiveness represents, with relation to snails, the following describes arrangement of documents, such as emails, according to inclusiveness. In accordance with certain embodiments of the present invention, there is provided a method for arranging a large number of documents relative to one another. Documents can be any type of document, but a particular usage of the present invention relates to emails. Documents may be loosely described as including both “body text” and “metadata”. Body text refers to the viewable text of the document, while metadata (also referred to herein as the document “header”) includes information and fields concerning the document, for example the fields for: author, date of creation, subject, date modified, etc. With regard to emails, which are anticipated to be a particularly useful beneficiary of the present invention, a great deal of information is stored as metadata for each email, often including the above fields, and others, application specific, such as Conversation ID (in Microsoft Outlook). The header of each email displays some of the metadata for the email, for example header 11 includes From: and To: fields, as well as Date: (date sent) and Subject: fields.
Emails may very often be inclusive documents. As shown in
>Date: Sun, 25 July 2006 15:42:23 +0300
>From: Andy Anderson <Andy.email@example.com>
>Subject: Old House??
When a few documents from amongst a large number of documents relate to the same subject matter, it is convenient to have them grouped together. Various organizational techniques are known in the art, such as classifying documents according to metadata associated with the documents, for example according to creation date of the document, or the author's name, or the subject line. In accordance with certain embodiments of the present invention, there is provided a method for grouping documents according to included documents.
The method for grouping and arranging documents according to inclusiveness can be applied with at least two different outputs:
1) A set of directionally interlinked nodes are created, in which each node is associated with at least one document. Each document has header parameters and body text, and if more than one document exists with the substantially identical body text, all are associated with the same node. In order to determine if documents are substantially identical, normalization is applied to the text, to remove any irrelevant characters. The level of normalization is variable, and no normalization is also considered normalization, since it represents an equal output to input. Large degrees of normalization are also possible, as described below. As stated, identical documents are associated with the same node, and, in many cases, many documents will all be associated with the same node. The nodes are directionally interlinked as follows. A first node is arranged or indicated to be a descendant of a second node if the body text of each document associated with the first node is substantially inclusive of the body text of each document that is associated with the second node.
The directionally interlinked nodes can be displayed by a processor on a display unit as a representational tree of nodes. The representational tree shows graphically the relationship between the nodes, indicating the degree of inclusiveness that each document associated with the nodes has relative to the others in the tree. So, in effect, the nodes of the tree graphically display the order of the documents, starting with the shortest one and leading to the longest, most inclusive document.
2) A group of documents that all include a certain document text is created. From this group, certain inclusive documents are identified as being inclusive of the others. In accordance with certain embodiments of the method, there is an advantage in that a reviewer can read all the text content of the entire group of documents by reading only the identified documents.
These two outputs are provided for exemplary purposes only, and should by no means be considered as limiting the scope of the invention.
In both of the above cases, it is noted that inclusiveness includes inclusiveness of whole documents, wherein a whole document has a header as well as body text.
However, there are at least three cases in which documents are considered inclusive even though the body text of one is not wholly inclusive of the other. Firstly, normalization may have stripped the body text of part of the text. Secondly, if a first document is inclusive of a document that is near-duplicated to a second document, the first document is considered inclusive of the second document, and the node associated with the first document is a descendant of the second document. Thirdly, it may be desired to use only a portion of the body text for the purposes of determining inclusiveness. For example, one might want to consider only the first 2000 characters of all documents. Other variants besides the specified example are also applicable.
Both of these outputs will be described in more detail with reference to
A variety of methods may be used in order to group and arrange documents according to inclusiveness; the present disclosure will describe example methods, from which other equivalent methods can be extrapolated.
Documents and emails often indicate that other, previous, documents once existed. An example of this in
In one manner of implementation of the present invention, each document that is to be classified is first provided with identification. This identification is used in accordance with certain embodiments and is not required in other embodiments of the present invention; identification is a method that will be referred to later in this disclosure. The present example is used with reference to emails, but may be used with other types of documents.
A large number of emails are provided, so that they may be organized according to inclusiveness. Each email is first provided with three types of identification: an email number, (MailID), an index value (Idx), and an individual, unique document number (DocID). These may be entered into the data structure. The MailID is assigned according to the number of emails that have previously been processed, the index value is set at zero, and the DocID is identical to the MailID. The purpose of these latter two identifiers will be explained in greater detail below.
Emails, as mentioned, in many cases, are expandable into a set of presumed, or derived or secondary documents. With reference now to
Presumed emails may be derived based on the appearance of header type text within the body text. For example, Body text bX0 contains several instances of header-type text. Header-type text indicates that all the remaining text in the email, after the header-type text, is likely to have been originally the body text of an earlier document. Thus a presumed email can be created for each instance of header-type text. The first instance of header-type text indicates a presumed email, and this presumed email is provided with identification as follows: MailID remains X, DocID is a numeric string specific to this particular presumed email, and the index value is 1, implying that this is the first presumed email originating from the original email X0. This first ‘presumed document’ is referred to herein as document X1, and it consists of header hX1 and body text bX1. Body text bX1 is derived from a copy of all the remaining body text following the header-type text identified as header hX1 in document X0.
The remaining text of body text bX0, following the header-type text already located, can still be searched for further instances of header-type text, to create further presumed emails. A second instance of header-type text is found, to suggest a second presumed email, with identification as follows: MailID remains X, DocID is a numeric string specific to this particular presumed email, and the index is 2, implying that it is the second presumed email originating from document X0. This second ‘presumed email is referred to herein as X2, and includes header hX2, composed of the header-type text that caused it. All body text that follows header hX2, namely body text bX2, forms the body text for presumed email X2.
The remaining text of body text bX0 is also expandable, and a third instance of header-type text can be located. This forms presumed email X3, with header hX3 formed of the header-type text that indicated the presumed email, and with body text formed of all of body text that follows header hX3, namely body text bX3.
With reference now to
As has been shown, it is possible to ‘expand’ an email or other document into an email plus a set of presumed emails that were included within the body text. The set of presumed emails can then be used for comparison with other emails to determine which other emails should be considered in the same group. However, creating ‘presumed documents’ as described is only one way of implementing the present invention, and the organization of documents according to inclusiveness may be alternatively implemented in many other ways.
According to certain embodiments of the invention, documents whose contents have not been derived from within the body text of other documents of the set may be referred to herein as ‘original documents’, this term is relative to ‘presumed documents’ that have been derived from others. ‘Original documents’ however need not be true document originals, for example they may be extracted from bulk files or document copies.
Having described ‘presumed documents’ in accordance with certain embodiments, the following seeks to describe the process of preparing fingerprints (for example, CRC, implemented using MD5) of the documents. The fingerprints are prepared for the purpose of comparing original and presumed documents with other documents from the group of documents. Two documents that have the same fingerprints, have with high probability the same content.
In order to determine whether documents should belong to the same group—that one is inclusive of the other, or that two are identical to one another, a document comparison may be done. “Matching” and “equivalent” and “duplicating” are other terms that refer herein to two or more texts that are substantially identical.
According to one embodiment, a copy of each document that has been processed is stored in a repository, and as each new document is processed, it is compared with previously stored documents, before being stored itself. According to one embodiment, the repository is part of the data structure, or it may be a separate unit.
One implementation involves the following: the document in its entirety is not used for the comparison since it may be huge. Instead, only a fingerprint of the text of each document is compared. The fingerprint is created as follows. A first fingerprint, for example, a CRC (cyclic redundancy check), or for example CRC-MD5, is compiled for the body text or the normalized body text of the document. A subject parameter is determined for the document, normalized and a second CRC is compiled for the normalized subject parameter. If the document is an email, the subject parameter may be derived from the Header parameter of the field labeled “Subject:”. For example, in
Fingerprints for the four documents X0, X1, X2 and X3, of
Having described the compilation of fingerprints for the documents, the process of normalization of text is described.
In one embodiment, before creation of the fingerprint, normalization is applied to the body text and/or subject parameter. The fingerprint would then be compiled from the normalized text. Normalization includes a 1:1 ratio, in which the normalization output is equivalent to the input, and such normalization is ineffective. In this case, the text may be referred to as normalized text, even if no normalization function has been applied to it. However, normalization can also remove features that are not directly part of the body text or subject. For example, for most documents, normalization can be applied to strip all formatting. This includes font, font size, italics, bold, color, etc. For email documents, normalization may also include removing characters that were added by the software document creation program. An example of added characters is the “>” sign typically added by email programs when a user replies to an email. The “>” sign at the beginning of a line indicates that that line of text is copied from the mail to which the user is replying. Normalization can also remove headers, footers, attachments and attachment notices, disclaimers, and signatures. Another example of added characters is in the subject parameter of headers. When replying to emails, email programs typically add to the subject line the characters “Re:”, and when forwarding, they add the characters “Fw:” or “Fwd:”. Other characters that are sometimes added to the body text, by email programs include “=20”, added when incorrectly converting from certain email programs to others. Sometimes a vertical line is added on the left to indicate included text from a previous email. Superfluous spaces are sometimes added, tabs, new line, etc. For the purpose of comparison between documents, normalization is applied to the body text and/or subject parameter. However, the original body text and subject parameter text is also preserved.
According to certain embodiments, the presence of signatures, disclaimers, anti-virus lines, free mail program notification (program added text), and attachment notification, can result in the lack of detection of true duplicates. For example, two emails may seem different only because they contain different disclaimers.
Alternatively, two emails may seem to be near-duplicates, when really they are very different, but they contain the same long disclaimer. It has been described above that these parameters may be removed in the normalization process. According to certain embodiments, a further category of normalization is described here. Disclaimers and the like are stripped from the document, and are replaced with a single word or number, unique to each type of disclaimer, signature, etc. This action reduces the influence of a long disclaimer on the rest of the text. Thus two emails whose main texts are not near duplicates will not suddenly appear as near-duplicates due to having identical disclaimers. Yet also two emails that have identical texts with different disclaimers, are related but are not true duplicates, so the presence of the replacement word ensures that the text analysis categorizes them as near-duplicates. According to one embodiment of a method for applying the invention, the method includes removing each of disclaimers, signatures, program added text and attachment notifications from the body text of documents, and replacing each unique disclaimer, signature, program added text and attachment notification with a unique short identifier prior to the creation of the document fingerprint.
In accordance with certain embodiments, there is provided a sequence of operations using normalized fingerprints, to determine equivalence and inclusiveness of documents. The embodiment should not be construed as limiting the invention in any way, as a number of alternative embodiments also exist. For example, other means can be implemented to assign values, or the entire system may use Object Oriented Programming instead of a data structure or database, etc. Data structure details can be organized differently, and may be organized in terms of trees instead of in terms of documents, as described herein. A non-limiting form of data structure is an SQL database, however, the invention does not need to use a data structure in any way. Hence, the following embodiment should not be construed as limiting, but is provided for exemplary purposes only.
Equivalence and inclusiveness may be alternatively accomplished by an individual algorithm, or another method may be used entirely to achieve the intent of the present invention. In one example, all documents can be first expanded into all of their component presumed documents, before comparison. Or, in another example, only the most inclusive documents are identified. Or, in a further example, selected documents from amongst a large group are processed. These selected documents may refer to all of the documents, or a defined subgroup from amongst the documents. Yet the following is a suggested embodiment. It is assumed that the document presented for processing has not yet been searched for header-type text, nor has any normalization yet been applied, nor have fingerprints been created.
The following is a specific example that serves to illustrate a broader concept. The specific example involves certain details that only represent one embodiment of the present invention, and is provided for exemplary purposes only. In accordance with certain embodiments of the invention, a processor is used with associated display, for communicating with the data structure. The processor is capable of manipulating the data of the data structure and displaying selected data on a display unit. With reference to
In step 301, the document is assigned a DocID value, and an index of zero. For example, the document may be assigned a DocID of 100.
In step 302, a field is set up for the document being processed, entitled MailID. This is initialized as the DocID of the document.
In step 303, the three identifiers described for the document are stored in a data structure, for example, for a first document, the particulars might be as follows:
In preliminary step 304 a, fields from the e-mail metadata are considered header fields. These include, but are not limited to the following fields: To, CC, BCC, Sent time, Internet Message ID, In Reply Message ID, Subject.
In step 304 b, the abovementioned fields from the e-mail header are parsed, including the fields: From, To, CC, BCC, Sent time, Internet Message ID, In Reply Message ID, Subject, Body, attachments. The parsed information is stored in a data structure.
In step 305, the Subject parameter is normalized and a CRC-MD5 is computed on the normalized subject (Nsubject). Both the subject and the CRC of the normalized subject are stored.
In step 306, a CRC on the normalized body text (Nbody) is calculated.
In step 307, the normalized body CRC and normalized subject CRC are compared with others in a repository for CRCs, and if a matching CRC pair already exists in the repository, the DocID of this document is linked to that of the match, and the document is finished being processed.
Otherwise, step 308, a loop, is accessed. This loop enables multiple presumed e-mails, from the body text of the original email, to be located, indexed, and processed. In step 308 a, the body text of the original email is searched to locate the next presumed email. In step 308 b, if a presumed email is found, the presumed email is provided with identification, and its index is set to one greater than that of the previously processed presumed email. In step 308 c, one is directed to follow again all the steps to process repeatedly the remainder of the body of the email for successive instances of header-type text, as stipulated in steps 303 onwards; for each instance, to construct a presumed document, index it, create a CRC pair, compare it for matching documents, and link if found. In step 308 d, when no more presumed emails can be located within the body text of the original email, the document is finished being processed.
The following table shows the index value that has been assigned to each of three presumed documents that are derived from the original document with DocID of 100.
Note that each document can be derived from the document of next lowest index value. For example, document 110 can be derived from 100 (as indeed it was), and document 120 from 110. Therefore document 100 is inclusive of document 110, and document 110 is inclusive of document 120. Each document is inclusive of the document of the next index value. Additionally, document 100 is inclusive of all documents that are equivalent to document 110. Inclusiveness of documents and equivalent documents is described with reference to
The following representation details the comparison of documents for equivalence, and also details the construction and comparison of presumed documents, the latter being similar to the above description with reference to
In the following representation, a number of documents are to be compared with one another for equivalence. The documents involved are original documents X0 and Y0 from
Fingerprints have been created for each of the documents, namely dX0, dY0, dP0 and dQ0 respectively. None of the documents have yet been reconstructed to find presumed documents within them. With reference now to
With reference now to
With reference to
With reference to
With reference to
With reference to
The following sequence of operations is presented in
With reference once again to
With reference now to
When a first presumed document is found within an original document, the first presumed document becomes the parent of the original document, because it is assumed to have happened before original document. When a second presumed document is found, it is assigned to be the parent of the first presumed document, since the second presumed document is more embedded than the first presumed document, and it is therefore assumed to have occurred earlier. The node associated with each sequentially derived presumed document (derived from the same document) becomes a parent of the node associated with the previously derived presumed document.
The root of a tree is associated with a node that has no parents. A document with no parent is an original or presumed document that does not include within its text any further presumed documents. In step 502, a field is filled in (for each document), in the data structure, entitled ParentNode. This ParentNodeID indicates which other document is the adjacent node in the root direction on a tree. The root of the tree is the node associated with least inclusive document of the tree—the document presumed to have been the earliest. Initially, ParentNodeID is set for all documents at −1, indicating that the document should be associated with a node having no parents. However, when a parent node is identified for a document, the field of ParentNodeID is updated with the NodeID of the parent node. In the example, the ParentNodeID field for Document 130 is identified as −1. All the other documents take their ParentNodeID according to the document index as reflected in their NodeID. The following table shows the ParentNodeID for each of the example documents:
In step 503, a field is filled in (for each document), entitled TreeID. This field identifies all documents that should belong to the same tree with the same value. Initially, TreeID for all documents is set as a copy of the NodeID field of the document. Beginning with a document that has a value for ParentNodeID of −1, the TreeID for that document remains as the NodeID. Descendants, or children, of that parent node, as indicated by their ParentNodeID field, then have their TreeID updated to that of their parent node. Then, the next generations of children have their TreeID updated to that of their parent node. In this way, all documents belonging to the same tree have the same node. The following table shows the TreeID for each of the example documents:
In step 504, a field can be filled in (for all documents) entitled UNION EQUISET. This field relates to linking trees that are identified as belonging to document sets, to be displayed together, as will be described with reference to
It is also possible to merge trees that are indicated as similar by the field of UNION EQUISET. In a practical example, if a reviewer wishes to view two documents that are near duplicates of one another, together he may be able to merge the two trees that they are each part of. The two nodes may be placed adjacent to one another, with or without hyperlinks, and time rest of the trees to which they each belong are similarly put adjacent to or merged with each other, as part of the same tree.
In step 505, for documents that have the same TreeID, the lowest value of NodeID is used as an update value for the TreeID. As mentioned before, using the lower value for the update ensures that earlier documents do not get updated, and remain with their assigned values. The following table shows the updated TreeID for each of the example documents:
As can be seen from the above table, all the documents have a TreeID of 100, and therefore are in the same tree.
Email messages often contain peripheries, such as disclaimers, signatures, and attachments. In one embodiment of the present invention these are suppressed prior to creating a fingerprint and comparing fingerprints. However, in another embodiment, they may be included in the CRC, or a CRC for these may be used as well. In addition, even if not used for purposes of document creation, these peripheries may be affiliated with the node containing the document to which they pertain, for a reviewer to access.
The following description relates to certain embodiments of the present invention, in which there is provided a method to arrange the aforementioned documents and presumed documents into groups according to their degree of inclusiveness, and according to their equivalence with other documents. Generally speaking, the groups are in the form of trees, as will now be described with reference to
However, the aforementioned documents can alternatively be arranged in the form of columns or lists, or simply individual documents can be selected for perusal, as will be described with reference to
With reference to
By way of example, if the documents of
As shown in
The tree is designed to organize the documents into linked nodes. Each node is associated with a group of documents that are equivalent, or at least substantially so (for example, they may have characters that are removed in the normalization of the subject procedure). Links between nodes imply that the text of a document associated with a node on the leaf side of the link includes the text of a document associated with a node on the root side of the link. For example, the text of the document associated with node 86 includes the text of the document associated with node 85.
According to certain embodiments of the present invention, the data structure is arranged in the form of trees, and a tree is defined as a set of directionally interlinked nodes. Each node is associated with at least one document, and all documents associated with a given node have substantially the same body text. Also, all documents having substantially the same body text are associated with the same node. The directional interlinkage indicates that if a first node is a descendent of a second node, the body text of each document associated with the first node is substantially inclusive of the body text of each document that is associated with the second node. For example, node 86 is a descendant of node 85, and the document associated with node 86 is substantially inclusive of the body text of any and all documents that are associated with node 85. The second node is a parent node of the first node. In the example just given, node 85 is termed the parent node of node 86.
The root, node of a tree (83) is the ultimate parent of the tree, having no parents itself. In different trees, the root node may have one or more descendant nodes, (descendants include intermediate and root nodes) or a root may have no descendants. In
With reference now to
In one embodiment, presumed documents are ‘put into’ or associated with the same tree as the original document from which they are derived, the root node is the smallest presumed document, that is, the last one to be derived from the original document. The other presumed documents are arranged in the root direction (ie, towards the root) according to their degree of inclusiveness. As a result, a node is associated with a document that has all the body text of all the other nodes in the direction of the root. In addition, documents that are equivalent, that is, they have the same body text and subject line, as determined by the document fingerprints, share a node. In this way, more complex trees can be created than simply by stringing together trees according to their original composition within a document. It is to be noted that in another embodiment, presumed documents are not associated with nodes, they are only used in order to determine the nodes that other documents, that are equivalent to the presumed documents, should be associated with. However, in a farther embodiment, also presumed documents are associated with nodes, sometimes forming a ‘missing link’ when no original email is equivalent to them.
In the present example shown in
In one embodiment, root node 92 links to, (also termed: “is affiliated with” or otherwise displays) the body text and subject line (non-normalized) of document X3. Root node 92 is also affiliated with certain header parameters of document X3, taken from the header or metadata of the document, and arranged in a table for easy perusal. First intermediate node 93 (adjacent root node 92) contains the body text and subject line of document X2, and is affiliated with header parameters from document X2. Second intermediate node 94 contains the body text and subject line of document Y0, and is affiliated with header parameters of both documents Y0 and X1. Leaf node 95 contains the body text and subject parameter of document X0 and is affiliated with header parameters of document X0. The header parameters may be affiliated by means of hyperlinks to the original document, or the header parameters may be copied into a hyperlinked or otherwise affiliated table.
It will be noted that only two of the nodes are associated with original documents, that is, second intermediate node 94, associated with document Y0, and leaf node 95, associated with document X0. The fact that a single tree includes nodes associated with two different original documents, namely Y0 and X0, indicates to a reviewer that the contents of the two documents X0 and Y0 are related. Since they occur along the same thread within the tree, the reviewer can see that document X0 contains all of the text of document Y0 as well as some additional matter.
With reference now to
In accordance with certain embodiments of the present invention, additional features of a tree and its construction relate to the fact that emails can be linked into a tree even if they have different subject parameters from one another. The factors that determine if nodes are joined to form a tree include descent, and equivalence of the associated document. It is possible for a user to have changed the subject line of an email in the course of a correspondence. However, if a more inclusive email exists, this can be expanded into a selection of presumed emails, in which the earlier dated ones will have the first subject line, and the later dated ones will have the second subject line. The presence of the more inclusive email indicates that although the subject line has changed, the documents are nevertheless part of a single conversational thread. The more inclusive email is expanded into a set of presumed documents that remain indexed together. As a result, the presumed documents—and all the documents that duplicate to them, are joined into the same tree. Thus, although the subject line of some of these documents is one thing, and the subject line of other of these documents is another, they are nevertheless presented on the same tree, as they were part of the same thread, as indicated by the more inclusive document's expanded contents.
Another benefit of an embodiment of the tree presentation of documents is that a reviewer probably does not wish to read through duplicated documents twice. He can read just one copy or just the subject and body text of a document associated with the node, in response to clicking on a node, if the node is represented on a display as a clickable icon. Or if the node is not a clickable icon, it may give other written directions or a different form of affiliation to the document text. In addition, the reviewer may still have access via the node to a hyperlink to metadata of all the original equivalent documents. If the node is associated with only a presumed document, the node can affiliate to metadata from the presumed documents too (or the node may simply be suppressed from view, as preferred).
In accordance with certain embodiments of the present invention, header parameters of equivalent documents grouped into a table and accessed from the associated node make it particularly easy for a reviewer to compare header parameters and/or statistics between the different copies of ostensibly the same document. For example, he may see which copy is recorded as having been sent first, find out if there are date discrepancies (perhaps one copy of the same email was sent on different days to different people), and find out who recipients of documents are, and more. All this information is available and can even be presented as a list or a table for easy comparison between parameters. Yet in some instances, this information is not required. Therefore, in a further embodiment of the present invention, all metadata is hidden from casual view and presented only upon clicking on an icon or similar reference.
In some cases, a reviewer wishes to have access to all body texts and subject lines, but does not wish to spend time reading anything twice. For example, if the contents of one document are all included within a second document, the reviewer may prefer to be able to read only the inclusive document. Certain embodiments of the present invention provide that documents are organized and presented according to trees, as described above, enabling the reviewer to selectively read only the leaf node, and he will be assured that he has accessed all the content for the documents associated with all nodes leading to that leaf. Additionally, a processor may be configured to indicate on a display unit which nodes are the leaf nodes. For example, leaf nodes may be marked or highlighted, so that the reviewer should know which to read. Additionally, in response to a user command, the processor can be configured to mark nodes for the display unit, in order to indicate whether a thread has been read, the relevance or priority of the thread, or the level of importance of the thread. The processor may further be configured to allow reviewer comments to be added to the display unit.
An additional embodiment of the present invention includes a ‘Leaf Compare’ tool—that is, a text comparison tool that compares between the text of the emails associated with two different nodes, for example between two leaves of the same tree. Using the leaf compare tool, a reviewer can access just the differences between two leaves, without having to retread the content common to both leaves. The text compare tool may be similar to standard document compare tools, or application specific. It preferably enables three different text formats, indicating deleted text (that is, text from a first node that does not appear in a second node), added text (that is, text from a second node that does not appear in a first node), and common text (that is, common text to both nodes), between two (or more) leaves. Examples of different text formats include but are not limited to, red strikeout text for deleted text, blue underlined text for added text, and plain black text for common text.
In accordance with a further embodiment of the present invention, the presentation of the tree may additionally indicate which nodes are associated with original documents, and which nodes are associated with presumed documents. For example, nodes associated with only presumed documents may be colored a different color, italicized or grayed. In a different embodiment, a node that is associated with both a presumed document and also an original document will only display an affiliation with the original document, while the affiliation to the presumed document is suppressed from view. In this way, readability is enhanced.
In a further embodiment, the whole tree structure is suppressed from view (or is not created). In this case, groups of documents that each contain a common presumed document are identified. This identification can be made using the algorithms disclosed in this disclosure, or another method. The documents from the group that are the most inclusive are identified and displayed for review. For the purposes of this embodiment, the relationship between the other documents does not need to be determined. This is described with reference to
In a further embodiment of the present invention, documents are not only compared for equivalence but also for near-duplication. A suggested algorithm for determining near duplication maybe found in co-pending application: U.S. application Ser. No. 11/572,441, whose contents are incorporated herein by reference. Determining that documents are near-duplicates of one another, enables near-duplicate documents to be presented to a reviewer as such. This is very useful for a number of reasons. Firstly, documents that are near duplicates of one another usually refer to the same subject matter, so it is convenient to review them side by side. Secondly, documents that are near-duplicates very often actually begin as identical documents, and are changed in minor ways, by the user, or, inadvertently, by a computer program. Thirdly, the level of near duplication can be variable by a user, enabling the user to define how close to each other two documents need to be in order for them to qualify as near-duplicates, and for the user to have them presented together for review. Fourthly, near-duplicates can pick up the similarity between two documents when the second document is basically identical to the first, but has had added comments interspersed amongst the text. For example, in an email reply, sometimes a user adds a word like “yes” somewhere in the middle of the body text of the email. This additional word can prevent the two documents—the email reply, and the email to which it is replying—from being strung together on the same tree. Yet the process steps for near duplication can indicate that the rest of the text is basically identical, which can help a reviewer understand the relative context of the two emails.
The following represents the use of comparison for near duplication, in accordance with an embodiment of the present invention. In the process of comparing the documents for duplication, the documents are also compared for near-duplication. At least a portion of the body text of each document is compared for near-duplication with at least a portion of the body texts of other documents. The comparison enables the detection and indication of near-duplicated documents. If near-duplication is found, an association is created between the documents found to be near-duplicates of one another. In accordance with a further embodiment, nodes that are associated with documents that are near duplicates to one another are assigned to the same document set. All other nodes in the trees that contain the near-duplicated nodes, are similarly assigned to that document set, as will be shown with reference to
In accordance with certain embodiments, documents determined to be near-duplicates of one another, are not presented as the same node on the same tree, but are presented in combination, that is, in close proximity to one another on a display unit as shown, or otherwise electronically linked to one another. The link indicates the fact that the two trees each contain a node, that are associated with documents that are near-duplicates to one another. For example, a tree has a set of nodes which may be presented on the top section of a computer screen. A document that near-duplicates to any of the nodes of the tree is presented as an individual node directly below the node to which it near-duplicates. Sometimes, a document that near-duplicates to a document in a tree, is itself part of a different tree. In this case, the two trees can be both presented to a reviewer together, and are called a ‘document set’. With reference now to
Document sets are another example of data structure. Document sets may be sets of nodes or sets of documents according to their association with nodes. In accordance with an embodiment of the invention, a method for marking the documents as associated with a certain document set is as follows: a first document is associated with a document set; all documents that are associated with a node that is linked to the node associated with the first document (or another previously added document) are then also associated with the document set. All documents that near-duplicate to a document already in the document set are also associated with the document set.
In the present example, the related trees are connected by the fact that one or more nodes near-duplicate to each other between the two trees. The two trees are displayed in close proximity to one another. This is just one specific way of indicating that the nodes associated with near duplicate documents are affiliated with one another. This is only one form of representing the relationship between near duplicates and there are many other acceptable ways to represent them, considered within the scope of the present invention.
In a further embodiment of the present invention other forms of document sets are described, enabling two or more trees to be affiliated with one another, indicating that they probably contain related subject matter. In a case where the documents are emails, there are email programs that store message identifiers to identify individual emails with a string of characters. When a user presses Reply-to, or Forward, in order to construct his email based on a previous email, these email programs create the message identifier for the new email to include a reference to the previous email. By parsing the message identifier, it is possible to determine which emails are replies or forwarding of which other emails. These emails are likely to be related to one another, so such emails are grouped together in a document set. In other words, emails whose Message ID indicate that they are connected to one another are not strung together on the same tree, but the trees in which each of them are nestled are presented together on the screen.
In a further embodiment of the present invention, document sets are constructed based upon a Conversation ID. Conversation ID is a feature of Microsoft Outlook, and is similar to the Message identifier described in the previous paragraph. Basically a Conversation ID contains a string of characters identifying an email. The basic character string for a new email may be 44 bytes long. When a person replies to, or forwards the email, an additional 10 bytes are added to the 44 byte string. Thus it is easy to see which emails are replies to other ones. Documents identified as being related to one another based upon Conversation ID are presented as document sets. The fact that their subject matter is likely to be related is indicated to the reviewer by the presentation of two or more trees in close proximity, or linked to one another. In accordance with one embodiment, a method for the invention includes creating an association between nodes that are associated with documents having related Conversation ID or Message ID indicators. The documents may be displayed grouped according to document sets.
The above description describes the use of trees and nodes in representing duplicating documents, parent-child relationships, and near duplicates. The following description relates to different types of trees. In
As may be seen in
According to this embodiment, the data structure comprises one or more trees in which a tree comprises at least a trunk and at least one node. Each node is associated with a document, and the trunk is associated with (zero or more) documents. All documents whose body text includes the same included document are associated with the same tree, and each unique inclusive document, as well as documents that duplicate to this unique inclusive document, is associated with one of the nodes of the tree. Each included document, as well as documents that duplicate to included document, are associated with the trunk of the tree.
One method for producing the trunked tree is using the embodiments described above with respect to
In one embodiment, in order to presence header parameters associated with parent nodes, the header parameters of all the documents associated with the trunk are associated with the leaf nodes. In other words, if a document associated with the node has had a presumed document constructed from it, one can additionally affiliate header parameters from all the documents that matched the presumed documents with the node. This is because the nodes associated with the presumed documents have been suppressed from display.
In accordance with a further embodiment of the present invention, another form for outputting documents, that have been arranged according to inclusiveness, is provided, as opposed to presenting them as trees. The trunk or the nodes may be displayed with one of many forms of tree symbolism; yet display alternatives also exist, as will be shown. Documents associated with the trunk or the node are identified as being grouped according to their association with the trunk or tree, but do not have to be presented in terms of the group.
As shown in
In another example, shown in
In accordance with another embodiment of the present invention, another form for outputting documents, that have been arranged according to inclusiveness, is provided, as opposed to presenting them as graphical trees. For example, documents may simply be listed in the data structure in sequential order and be supplied with a column listing indicating with which tree, and with which section of the tree, they are associated. As shown in the first column of
In another example, the columns could represent trunked trees as shown in
There have been described above, many non-limiting different ways in which documents can be arranged according to duplication, inclusiveness and, according to certain embodiments, according to document sets. The following seeks to describe document properties and document sources that can be used in accordance with embodiments of the invention.
Another benefit of an embodiment of the present invention is its ability to have access to files such as emails from different computers, or from different email archives, for example different PST files. According to certain embodiments, the operation of the invention is not to immediately assign documents extracted from different bulk files to different trees, but rather assigns documents to trees according to inclusiveness, irrespective of which file the document originated from. Thus a tree may be composed of documents originating from a number of different trees. According to further embodiments, it is also irrelevant whether the documents were in whole document format or needed extraction to document format before the invention is applied. Similarly, the invention is not limited to only associating documents with nodes on the same tree if the subject parameter for all the documents is the same. On the contrary, for different nodes, the subject parameter may be different, but as long as the documents satisfy the inclusiveness condition, they may be associated with nodes of the same tree.
As mentioned before, according to certain embodiments, a node can even be associated with a “presumed document”—or a “presumed email”—an document that has not been extracted from either of the PSTs 134 or 136, but is nevertheless presumed to have existed once, since a header for it, and text, is included within the text of a document that is found in one of the PSTs.
As mentioned, Tree 140 includes a number of nodes 142,and each node is associated with one or more documents. Each document is associated with its own node, unless the document duplicates to another document. If the document is a duplicate of another document, both of them are associated with the same node. Each node forms its own tree, unless it is a descendant of a different node. From the contents of more than one PST, it is expected that a high number of different trees will be used to organize the documents. A node belongs to a tree if it is a descendant of another node in the tree.
It will also be understood that the system according to the invention may be a suitably programmed computer. Likewise, the invention contemplates a computer program being readable by a computer for executing the method of the invention. The invention further contemplates a machine-readable memory tangibly embodying a program of instructions executable by the machine for executing the method of the invention.
Many specificities have been provided in the above description, however, these should not be construed as limiting the invention in any way. In addition, certain details have not been described, as they would be obvious to persons skilled in the art.
The invention should be understood in terms of the appended claims.
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7899831 *||25 Mar 2008||1 Mar 2011||International Business Machines Corporation||Method and system for folder recommendation in a file operation|
|US8359592||19 Mar 2009||22 Jan 2013||Microsoft Corporation||Identifying groups and subgroups|
|US8402096 *||24 Jun 2008||19 Mar 2013||Microsoft Corporation||Automatic conversation techniques|
|US8589426 *||29 Oct 2008||19 Nov 2013||Sprint Communications Company L.P.||Simultaneous file editor|
|US8725687||2 Apr 2013||13 May 2014||Commvault Systems, Inc.||Systems and methods for byte-level or quasi byte-level single instancing|
|US8788500||10 Sep 2010||22 Jul 2014||International Business Machines Corporation||Electronic mail duplicate detection|
|US8798366||28 Dec 2010||5 Aug 2014||Amazon Technologies, Inc.||Electronic book pagination|
|US8898177||10 Sep 2010||25 Nov 2014||International Business Machines Corporation||E-mail thread hierarchy detection|
|US8930361 *||3 May 2011||6 Jan 2015||Nokia Corporation||Method and apparatus for cleaning data sets for a search process|
|US8990088 *||28 Jan 2009||24 Mar 2015||Microsoft Corporation||Tool and framework for creating consistent normalization maps and grammars|
|US9015080||16 Mar 2012||21 Apr 2015||Orbis Technologies, Inc.||Systems and methods for semantic inference and reasoning|
|US9046983||12 May 2009||2 Jun 2015||Microsoft Technology Licensing, Llc||Hierarchically-organized control galleries|
|US9058117||9 Oct 2013||16 Jun 2015||Commvault Systems, Inc.||Block-level single instancing|
|US9069767||17 Mar 2011||30 Jun 2015||Amazon Technologies, Inc.||Aligning content items to identify differences|
|US9098473||4 May 2012||4 Aug 2015||Microsoft Technology Licensing, Llc||Accessing an out-space user interface for a document editor program|
|US9098837||9 Feb 2008||4 Aug 2015||Microsoft Technology Licensing, Llc||Side-by-side shared calendars|
|US20040268270 *||26 Jun 2003||30 Dec 2004||Microsoft Corporation||Side-by-side shared calendars|
|US20050005235 *||21 May 2004||6 Jan 2005||Microsoft Corporation||Adaptive multi-line view user interface|
|US20050005249 *||30 Apr 2004||6 Jan 2005||Microsoft Corporation||Combined content selection and display user interface|
|US20100070584 *||12 Sep 2008||18 Mar 2010||International Business Machines Corporation||Method and system for dynamic e-mail view conversion|
|US20100191519 *||28 Jan 2009||29 Jul 2010||Microsoft Corporation||Tool and framework for creating consistent normalization maps and grammars|
|US20110270606 *||3 Nov 2011||Orbis Technologies, Inc.||Systems and methods for semantic search, content correlation and visualization|
|US20120060082 *||2 Sep 2010||8 Mar 2012||Lexisnexis, A Division Of Reed Elsevier Inc.||Methods and systems for annotating electronic documents|
|US20120254202 *||4 Oct 2012||Nokia Corporation||Method and apparatus for cleaning data sets for a search process|
|US20120271793 *||26 Jun 2012||25 Oct 2012||Parag Gokhale||Application-aware and remote single instance data management|
|U.S. Classification||1/1, 707/E17.044, 707/999.101|
|Cooperative Classification||G06K9/00469, G06K9/00483, G06F17/30705, G06F17/2211, H04L51/16|
|2 Jan 2008||AS||Assignment|
Owner name: EQUIVIO LTD., ISRAEL
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:RAVID, YIFTACH;MILO, AMIR;REEL/FRAME:020319/0854
Effective date: 20071231