Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberCA2773219 C
Publication typeGrant
Application numberCA 2773219
PCT numberPCT/US2010/043383
Publication date15 Sep 2015
Filing date27 Jul 2010
Priority date28 Jul 2009
Also published asCA2773159A1, CA2773159C, CA2773194A1, CA2773194C, CA2773198A1, CA2773198C, CA2773219A1, CA2773220A1, CA2773220C, CA2773263A1, CA2773263C, CA2773317A1, CA2773317C, CA2773319A1, CA2773319C, EP2460096A2, EP2460097A1, EP2460098A1, EP2460099A1, EP2460100A1, EP2460101A2, EP2460102A2, EP2460103A1, US8515957, US8515958, US8572084, US8635223, US8645378, US8700627, US8713018, US8909647, US9064008, US9165062, US9336303, US9477751, US9542483, US9679049, US20110029525, US20110029526, US20110029527, US20110029529, US20110029530, US20110029531, US20110029532, US20110029536, US20130339275, US20130339360, US20140059048, US20140136539, US20140156564, US20140229478, US20140236947, US20150095337, US20150286957, US20160253408, US20170046617, US20170116325, WO2011017064A2, WO2011017064A3, WO2011017065A1, WO2011017080A1, WO2011017098A1, WO2011017133A2, WO2011017133A3, WO2011017133A4, WO2011017134A1, WO2011017134A4, WO2011017152A2, WO2011017152A3, WO2011017152A4, WO2011017155A1, WO2011017155A4
Publication numberCA 2773219, CA 2773219 C, CA 2773219C, CA-C-2773219, CA2773219 C, CA2773219C, PCT/2010/43383, PCT/US/10/043383, PCT/US/10/43383, PCT/US/2010/043383, PCT/US/2010/43383, PCT/US10/043383, PCT/US10/43383, PCT/US10043383, PCT/US1043383, PCT/US2010/043383, PCT/US2010/43383, PCT/US2010043383, PCT/US201043383
InventorsWilliam C. Knight, Nicholas I. Nussbaum
ApplicantFti Consulting, Inc., William C. Knight, Nicholas I. Nussbaum
Export CitationBiBTeX, EndNote, RefMan
External Links: CIPO, Espacenet
Displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor
CA 2773219 C
Abstract
A system (11) and method (40) for providing reference documents (14b) as a suggestion for classifying uncoded documents (14a) is provided. Reference electronically stored information items (14b) and a set of uncoded electronically stored information items (14a) are designated. Each of the reference information items are previously classified. At least one uncoded electronically stored information item (14a) is compared with the reference electronically stored information items (14b). One or more of the reference electronically stored information items (14b) similar to the at least one uncoded electronically stored information items (14a) are identified. Relationships are depicted between the at least one uncoded electronically stored information item (14a) and the similar reference electronically stored information items (14b) for classifying the at least one uncoded electronically stored information item (14a).
Claims(20)
What is claimed is:
1. A system (11) for providing reference documents (14b) as a suggestion for classifying electronically stored information (13) using nearest neighbor, comprising:
a clustering module (33) to provide a set of reference electronically stored information items (14b) that are each classified with a code and a different set of uncoded electronically stored information items (14a);
a similarity module (34) to compare at least one of the uncoded electronically stored information items (14a) from the different set with the set of the reference electronically stored information items (14b) and to identify one or more of the reference electronically stored information items (14b) that are similar to the at least one uncoded electronically stored information item (14a);
a processing module to process the classification codes associated with the similar reference electronically stored information items, comprising:
a type module to determine a number of different types of the classification codes associated with the similar reference electronically stored information items;
a presence module to determine one or more of a presence and absence of the similar reference electronically stored information items with each type of the classification codes; and a quantity module to determine for each type of the classification codes a quantity of the similar reference electronically stored information items;
a suggestion module to display a visual classification suggestion based on at least one of the presence and the absence and the quantity and the number of the types via a display of the at least one uncoded electronically stored information item and the similar reference electronically stored information items; and a receipt module to receive a classification code of one of the types for the at least one uncoded electronically stored information item from a human reviewer based on the suggestion.
2. A system (11) according to Claim 1, further comprising:
a reference set module (36) to generate the set of reference electronically stored information items (14b), comprising at least one of:

a comparison module to obtain a set of electronically stored information items (14a), to identify one or more electronically stored information items (14a) that are dissimilar from each other electronically stored information item, and to assign a classification code to each of the dissimilar electronically stored information items (14a), as the reference electronically stored information items (14b); and a reference clustering module to group electronically stored information items (14a) for a document review project into one or more clusters (72), to select one or more of the electronically stored information items (14a) in at least one cluster (72), and to assign a classification code (76) to each of the selected electronically stored information items (14a), as the reference electronically stored information items (14b).
3. A system (11) according to Claim 1, further comprising:
a score module to form a score vector for each uncoded electronically stored information item (14a) and each reference electronically stored information item (14b); and the similarity module to calculate the similarity metric by comparing the score vectors for the uncoded electronically stored information items (14a) and the reference electronically stored information items (14b).
4. A system (11) according to Claim 1, wherein the similarity metric is calculated according to the following equation:
where cos .sigma.-AB comprises a similarity between uncoded electronically stored information item A and reference electronically stored information item B, ~A comprises a score vector for uncoded electronically stored information item A, and ~B comprises a score vector for reference electronically stored information item B.
5. A system (11) according to Claim 1, further comprising:
a generation module to automatically generate an additional classification code (76) for the at least one uncoded electronically stored information item; and an assignment module to assign one of the automatically generated classification code and the classification code received from the user to the at least one uncoded reference electronically stored information item.
6. A method (40) for providing reference documents (14b) as a suggestion for classifying electronically stored information (13) using nearest neighbor, comprising the steps of:
designating a set of reference electronically stored information items (14b) that are each classified with a code and a different set of uncoded electronically stored information items (14a);
comparing at least one of the uncoded electronically stored information items (14a) from the different set with the set of the reference electronically stored information items (14b);
identifying one or more of the reference electronically stored information items (14b) that are similar to the at least one uncoded electronically stored information item (14a);
processing the classification codes associated with the similar reference electronically stored information items, comprising:
determining a number of different types of the classification codes associated with the similar reference electronically stored information items;
determining one or more of a presence and absence of the similar reference electronically stored information items with each type of the classification codes; and determining for each type of the classification codes a quantity of the similar reference electronically stored information items;
displaying a visual classification suggestion based on at least one of the presence and the absence and the quantity and the number of the types via a display of the uncoded electronically stored information item and the similar reference electronically stored information items; and receiving a classification code of one of the types for the at least one uncoded electronically stored information item from a human reviewer based on the suggestion, wherein the steps are performed by a suitably programmed computer.
7. A method (40) according to Claim 6, further comprising:
generating the reference electronically stored information items (14b) from a set of electronically stored information items (14a), comprising at least one of:
identifying the electronically stored information items (14a) that are dissimilar from each other electronically stored information item and assigning a classification code (76) to each of the dissimilar electronically stored information items (14a), as the reference electronically stored information items (14b); and grouping a set of electronically stored information items (14a) associated with a document review project into one or more clusters (72), selecting one or more of the electronically stored information items (14a) in at least one cluster (72), and assigning a classification code (76) to each of the selected electronically stored information items (14a), as the reference electronically stored information items (14b).
8. A method (40) according to Claim 6, further comprising:
forming a score vector for each uncoded electronically stored information item (14a) and each reference electronically stored information item (14b); and calculating a similarity metric by comparing the score vectors for the uncoded electronically stored information items (14a) and the reference electronically stored information items (14b) in the reference set.
9. A method (40) according to Claim 8, wherein the similarity metric is calculated according to the following equation:
where cos .sigma.-AB comprises a similarity between uncoded electronically stored information item A and reference electronically stored information item B, ~A comprises a score vector for uncoded electronically stored information item A, and ~B comprises a score vector for reference electronically stored information item B.
10. A method (40) according to Claim 6, further comprising:
automatically generating an additional classification code (76) for the at least one uncoded electronically stored information item; and assigning one of the generated classification code and the classification code received from the user to the at least one uncoded reference electronically stored information item.
11. A system (11) for identifying reference documents (14b) for use in classifying uncoded documents (14a), comprising:
a database (30) to store a set of reference documents (14b) that are each classified with a code (76);
a clustering module (33) to designate a set of clusters (72) each comprising uncoded documents (14a) from a different set than the reference documents;
a similarity module (34) to select at least one of the uncoded documents (14a) from the different set, to compare the at least one uncoded document (14a) with each of the reference documents (14b), and to identify one or more reference documents (14b) that satisfy a threshold of similarity with the at least one uncoded document (14a);
a processing module to process the classification codes associated with the similar reference documents, comprising:
a type module to determine a number of different types of the classification codes associated with the similar reference documents;
a presence module to determine one or more of a presence and absence of the similar reference documents with each type of the classification codes; and a quantity module to determine for each type of the classification codes a quantity of the similar reference documents;
a suggestion module to display a visual classification suggestion based on at least one of the presence and the absence and the quantity and the number of the types via a display of the at least one of the uncoded document and the similar reference documents;
and a classification module to receive a classification code of one of the types for the at least one uncoded document from a human reviewer based on the suggestion.
12. A system (11) according to Claim 11, further comprising:
a reference set module to generate the set of reference documents (14b), comprising at least one of:
a reference similarity module to obtain a set of documents (14b), to identify the documents (14b) that are dissimilar from each other document (14a), and to assign a classification code (76) to each of the dissimilar documents (14b); and a reference cluster module to generate clusters (72) of documents (14b) for a document review project, to select one or more of the documents (14b) in at least one of the clusters (72), and to assign a classification code (76) to each of the documents (14b).
13. A system (11) according to Claim 11, wherein the one or more reference documents (14b) are selected from at least one of a predefined, customized, or arbitrary reference document set.
14. A system (11) according to Claim 11, wherein the similarity module comprises:
a score module to form a score vector for each uncoded document (14a) and each reference document (14b); and a vector similarity module to calculate a similarity metric between the score vectors for the uncoded documents (14a) and reference documents (14b).
15. A system (11) according to Claim 11, wherein the similarity metric is calculated according to the following equation:
where cos .sigma.AB comprises a similarity between uncoded document A and reference document B, ~A comprises a score vector for uncoded document A, and ~B comprises a score vector for reference document B.
16. A method (40) for identifying reference documents (14b) for use in classifying uncoded documents (14a), comprising the steps of:
designating a set of reference documents (14b) that are each classified with a code (76);
designating a set of clusters (72) each comprising uncoded documents (14a) from a different set than the reference documents;
selecting at least one uncoded document (14a) from the different set and comparing the at least one uncoded document (14a) with each of the reference documents (14b);
identifying one or more reference documents (14b) that satisfy a threshold of similarity with the at least one uncoded document (14a); and a processing module to process the classification codes associated with the similar reference documents, comprising:
determining a number of different types of the classification codes associated with the similar reference documents;
determining one or more of a presence and absence of the similar reference documents with each type of the classification codes; and determining a quantity of the similar reference documents for each type of the different classification codes;
displaying a visual classification suggestion based on at least one of the presence and the absence and the quantity and the number of the types via a display of the at least one of the uncoded document and the similar reference documents; and receiving a classification code of one of the types for the at least one uncoded document from a human reviewer based on the suggestion, wherein the steps are performed by a suitably programmed computer.
17. A method (40) according to Claim 16, further comprising:
generating the set of reference documents (14b), comprising at least one of:
obtaining a set of documents (14b), identifying the documents (14b) that are dissimilar from each other document (14a), and assigning a classification code (76) to each of the dissimilar documents (14b); and generating clusters (72) of documents (14b), selecting one or more of the documents (14b) in at least one of the clusters (72) and assigning a classification code (76) to each of the documents (14b).
18. A method (40) according to Claim 16, wherein the one or more reference documents (14b) are selected from at least one of a predefined, customized, or arbitrary reference document set.
19. A method (40) according to Claim 16, further comprising:
forming a score vector for each uncoded document (14a) and each reference document (14b); and calculating a similarity metric between the score vectors for the uncoded documents (14a) and reference documents (14b).
20. A method (40) according to Claim 16, wherein the similarity metric is calculated according to the following equation:
where cos .sigma.-AB comprises a similarity between uncoded document A and reference document B, ~A comprises a score vector for uncoded document A, and ~B comprises a score vector for reference document B.
Description  (OCR text may contain errors)

DISPLAYING RELATIONSHIPS BETWEEN ELECTRONICALLY STORED
INFORMATION TO PROVIDE CLASSIFICATION SUGGESTIONS
VIA NEAREST NEIGHBOR
TECHNICAL FIELD
This application relates in general to using documents as a reference point and, in particular, to a system and method for displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor.
BACKGROUND ART' Historically, document review during the discovery phase of litigation and for other types of legal matters, such as due diligence and regulatory compliance, have been.
conducted manually. Dating document review, individual reviewers, generally licensed attorneys, are assigned sets of documents for coding. .A reviewer must carefully study each document and categorize the document by assigning a code or other marker from a set of descriptive classifications, such as "privileged," "responsive," and "non-responsive." The classifications can affect the disposition of each document, including admissibility into evidence.
During discovery, document review can potentially affect the outcome of the .underlying legal matter, so consistent and accurate results are crucial. Manual document review is tedious and time-consuming. Marking documents is solely at the discretion of each reviewer and inconsistent results may occur due to .misunderstanding, time pressures, fatigue, or other factors.
A large volume of documents reviewed, often with only limited time, can create a loss of mental.
focus and a loss of purpose for the resultant classification. Each new reviewer also faces a steep learning curve to become familiar with the legal matter, classification categories, and review techniques.
Currently, with the increasingly widespread movement to electronically stored.
information (ESI), manual document review is no longer practicable. The often exponential growth of ESE exceeds the bounds reasonable for conventional manual human document review and underscores the need for computer-assisted ESI review tools.
Conventional ESI review tools have proven inadequate to providing efficient, accurate, and consistent results. For example, DiscoverReady LLC, a Delaware limited liability company, custom programs ES1 review tools, which conduct semi-automated document review through multiple passes over a document set in ESL. foam. During the first pass, documents are grouped by category and basic codes are assigned. Subsequent passes refine and further assign endings.
Multiple pass review requires a priori project-sped tic knowledge engineering, which is ODly useful for .the single project, thereby losing the benefit of any inferred knowledge or know-how for .use in other review projects.
Thus, there remains a need for a system and method tor increasing the efficiency of document review that bootstraps knowledge gained from other reviews while ultimately ensuring independent reviewer discretion.
DISCLOSURE OF THE INVENTION
Document review efficiency can be increased by identifying relationships between reference ESI and uncoded ESL., and providing a suggestion for classification based on the relationships. The uncoded ESI for a document review .project are identified and clustered. At least one of the uncoded .ESI is selected from the clusters and compared with the reference ESI
based on a similarity metric. The reference ESI most similar to the selected uncoded ES1 are identified.. Classification codes assigned to the similar reference ESI can be used to provide suggestions for classification of the selected uncoded. ESL Further, a machine-generated suggestion for classification code can be provided with a confidence level.
An embodiment provides a system and method for displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor.
Reference electronically stored information items and a set of unaided electronically stored information items are designated. Each of the reference information items are previously classified. At .least one uncoded electronically stored information item is compared with the reference electronically stored information items. One or more of the reference electronically stored information items similar to the at least one uncoded electronically stored. information items are identified. Relationships are depicted between the at least one uncoiled electronically stored information item and the similar reference electronically stored information items .for classifying the at least one uncoded electronically stored information item.
A further embodiment provides a system and method for identifying reference .documents :for use in classifying uncoded documents. A set of reference documents is designated. Each reference document is associated with a classification code. A set of clusters each including uncoded documents is designated. At least one uncoded document is selected and compared with each of the reference documents. One or more reference documents that satisfy a threshold of similarity with the at least one uncoded document is identified.
Relationships between the at least one .uncoded document and the similar reference documents are displayed based on the associated classification codes as suggestions for classifying the at least one uncoded document.

2

3 Still other embodiments of the present invention will become readily apparent to those skilled in the art from the following detailed description, wherein are described embodiments by way of illustrating the best mode contemplated for carrying out the invention.
As will be realized, the invention is capable of other and different embodiments and its several details are capable of modifications in various obvious respects. Accordingly, the drawings and detailed description are to be regarded as illustrative in nature and not as restrictive.
DESCRIPTION OF THE DRAWINGS
FIGURE 1 is a block diagram showing a system for displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor, in accordance with one embodiment.
FIGURE 2 is a process flow diagram showing a method for displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor, in accordance with one embodiment.
FIGURE 3 is a block diagram showing, by way of example, measures for selecting a document reference subset.
FIGURE 4 is a process flow diagram showing, by way of example, a method for comparing an uncoded document to reference documents for use in the method of FIGURE 2.
FIGURE 5 is a screenshot showing, by way of example, a visual display of reference documents in relation to uncoded documents.
FIGURE 6 is an alternative visual display of the similar reference documents and uncoded documents.
FIGURE 7 is a process flow diagram showing, by way of example, a method for classifying uncoded documents for use in the method of FIGURE 2.
BEST MODE FOR CARRYING OUT THE INVENTION
The ever-increasing volume of ESI underlies the need for automating document review for improved consistency and throughput. Previously coded documents offer knowledge gleaned from earlier work in similar legal projects, as well as a reference point for classifying uncoded ESI.
Reference documents are documents that have been previously classified by content and can be used to influence classification of uncoded, that is unclassified, ESI.
Specifically, relationships between the uncoded ESI and the reference ESI can be visually depicted to provide suggestions, for instance to a human reviewer, for classifying the visually-proximal uncoded ESI.

4 Complete ESI review requires a support environment within which classification can be performed. FIGURE 1 is a block diagram showing a system 10 for displaying relationships between electronically stored information to provide classification suggestions via nearest neighbor, in accordance with one embodiment. By way of illustration, the system 10 operates in a distributed computing environment, which includes a plurality of heterogeneous systems and ESI sources. Henceforth, a single item of ESI will be referenced as a "document," although ESI
can include other forms of non-document data, as described infra. A backend server 11 is coupled to a storage device 13, which stores documents 14a, such as uncoded documents, in the form of structured or unstructured data, a database 30 for maintaining information about the documents, and a lookup database 38 for storing many-to-many mappings 39 between documents and document features, such as concepts. The storage device 13 also stores reference documents 14b, which can provide a training set of trusted and known results for use in guiding ESI classification. The reference documents 14b are each associated with an assigned classification code and considered as classified or coded. Hereinafter, the terms "classified" and "coded" are used interchangeably with the same intended meaning, unless otherwise indicated.
A set of reference documents can be hand-selected or automatically selected through guided review, which is further discussed below. Additionally, the set of reference documents can be predetermined or can be generated dynamically, as the selected uncoded documents are classified and subsequently added to the set of reference documents.
The backend server 11 is coupled to an intranetwork 21 and executes a workbench suite 31 for providing a user interface framework for automated document management, processing, analysis, and classification. In a further embodiment, the backend server 11 can be accessed via an internetwork 22. The workbench software suite 31 includes a document mapper 32 that includes a clustering engine 33, similarity searcher 34, classifier 35, and display generator 36.
Other workbench suite modules are possible.
The clustering engine 33 performs efficient document scoring and clustering of documents, including uncoded and coded documents, such as described in commonly-assigned U.S. Patent No. 7,610,313. Clusters of uncoded documents 14a can be formed and organized along vectors, known as spines, based on a similarity of the clusters, which can be expressed in terms of distance. During clustering, groupings of related documents are provided. The content of each document can be converted into a set of tokens, which are word-level or character-level n-grams, raw terms, concepts, or entities. Other tokens are possible. An n-gram is a predetermined number of items selected from a source. The items can include syllables, letters, or words, as well as other items. A raw term is a term that has not been processed or manipulated. Concepts typically include nouns and noun phrases obtained through part-of-speech tagging that have a common semantic meaning. Entities further refine nouns and noun phrases into people, places, and things, such as meetings, animals, relationships, and various

5 other objects. Entities can be extracted using entity extraction techniques known in the field.
Clustering of the documents can be based on cluster criteria, such as the similarity of tokens, including n-grams, raw terms, concepts, entities, email addresses, or other metadata.
In a further embodiment, the clusters can include uncoded and coded documents, which are generated based on a similarity with the uncoded documents, as discussed in commonly-owned U.S. Patent Application Publication No. 2011/0029526, entitled "System and Method for Displaying Relationships Between Electronically Stored Information to Provide Classification Suggestions via Inclusion," filed July 9, 2010, pending, and U.S. Patent No.
8,515,957, entitled "System and Method for Displaying Relationships Between Electronically Stored Information to Provide Classification Suggestions via Injection," filed July 9, 2010.
The similarity searcher 34 identifies the reference documents 14b that are most similar to selected uncoded documents 14a, clusters, or spines, as further described below with reference to FIGURE 4. For example, the uncoded documents, reference documents, clusters, and spines can each be represented by a score vector, which includes paired values consisting of a token, such as a term occurring in that document, cluster or spine, and the associated score for that token.
Subsequently, the score vector of the uncoded document, cluster, or spine is then compared with the score vectors of the reference documents to identify similar reference documents.
The classifier 35 provides a machine-generated suggestion and confidence level for classification of selected uncoded documents 14a, clusters, or spines, as further described below with reference to FIGURE 7. The display generator 36 arranges the clusters and spines in thematic relationships in a two-dimensional visual display space, as further described below beginning with reference to FIGURE 5. Once generated, the visual display space is transmitted to a work client 12 by the backend server 11 via the document mapper 32 for presenting to a reviewer on a display 37. The reviewer can include an individual person who is assigned to review and classify one or more uncoded documents by designating a code.
Hereinafter, the terms "reviewer" and "custodian" are used interchangeably with the same intended meaning, unless otherwise indicated. Other types of reviewers are possible, including machine-implemented reviewers.

The document mapper 32 operates on uncoded 14a and coded documents 1 4b, which can be retrieved from the storage 13, as well as from a plurality of local and remote sources. The local sources include a local server 15, which is coupled to a storage device 16 with docunients .17 and a local client 18, which is coupled to a storage device 19 with documents 20. The local server 15 and local client 18 are interconnected to the backend server II and the work client .12 over an intranetwork 2.1. In addition, the document mapper 32 can identify and retrieve documents from remote sources over an imernetwork 2.2, including, the Internet, through a gateway 23 interfaced to the intranetwork 21.. The remote sources include a remote server 24, which is coupled to a storage device 2:5 with documents 26 and a remote client 27, which is coupled to a storage device 28 with documents 29. Other document sources, either local or remote, are possible.
The individual documents 17, 20, 26, 29 include all forms and types of structured and unstructured ESL including electronic message stores, word processing documents, electronic mail (email) folders, Web pages, and graphical or multimedia data.
Notwithstanding, the.
documents could be in .the form of structurally organized data, such as stored in a spreadsheet or database.
In one embodiment, the individual documents I4a, 1413, 17, 20, 26, 29 include electronic message folders storing email and attachments, such as maintained by the Outlook and Outlook Express products, licensed. by Microsoft Corporation, Redmond, WA. The database can be an SQL-based relational database, such as the Oracle database management system, Release 8, licensed by Oracle Corporation, Redwood Shores, CA.
The individual documents 17, 20, 26, 29 can be designated and stored as uncoded documents or reference documents. The uncoded documents, which are unclassified, are selected for a. document review project and stored as a document corpus for classification. The reference documents are initially uncoded documents that can be selected from the corpus or other source of uncoded documents, and subsequently classified. The reference documents can assist in providing suggestions for classification of the remaining uncoded documents based on visual relationships between the uncoded documents and reference documents..
In a further embodiment, the reference documents can provide classification suggestions for a document corpus associated with a related document review project. In yet a further embodiment, the reference documents can be used as a. training set to form machine-generated suggestions for classifying uncoded documents, as further described below with reference to FIGURE 7.
The document corpus for a document review project can be divided into subsets of uncoded documents, which are each provided to a. particular reviewer as an assignment. To

6 maintain consistency, the same classification codes can be used across all assignments in the document review project. Alternatively, the classification codes can be different for each assignment. The classification codes can be determined using taxonomy generation, during which a. list of classification codes can be provided by a reviewer or determined automatically.
For purposes of legal discovery, the list of classification codes can include "privileged,"
"responsive," or "non-responsive;" however, other classification codes are possible. A
"privileged" document contains information that is protected by a privilege., meaning that the document should not be disclosed or "produced" to an opposing party.
Disclosing a "privileged"
document can result in an unintentional waiver of the subject matter disclosed. A "responsive"
document contains information that is related to a legal matter on which the document review project is based and a "non-responsive" document includes information that is not related to the legal matter.
The system 10 includes individual computer systems, such as the backend server 11, work server 12, server 15, client 18, remote server 24 and remote client 27.
The individual computer systems are general purpose, programmed digital computing devices consisting of a.
central processing unit (CPU), random access memory (RAM), non-volatile secondary storage, such as a bard drive or CD ROM drive, network interfaces, and peripheral devices, including user interfacing means, such as a keyboard and display. The various implementations of the source code and object and byte codes can be held on a computer-readable storage medium, such as a floppy disk, hard drive, digital video disk (MD), random access memory (RAM), read-only memory (ROW and similar storage mediums. For example, program code, including software programs, and data are loaded into the RAM for execution and processing by the CPU and, results are generated for display, output, transmittal, or storage.
Identifying relationships between the reference documents and uncoded documents includes clustering and similarity measures. FIGURE 2 is a process flow diagram showing a method 40 for displaying relationships between electronically stored infomiation to provide classification suggestions via nearest neighbor, in accordance with one embodiment, A set of document clusters is obtained. (block 41), In one embodiment, the clusters can include uncoded documents, and in a further embodiment, the clusters can include uncoded and coded documents.
The clustered uncoded documents can represent a corpus of unaided documents for a document review project, or one or more assignments of uncoded documents. The document corpus can include all. -uncoded documents for a document review project, while, each assignment can include a subset of uncoded documents selected from the corpus and assigned to a reviewer. The corpus can be divided into assignments using assignment criteria, such as custodian or source of

7

8 PCT/US2010/043383 the -uncoded document, content, document type, and date. Other criteria are possible. Prior to, concurrent with, or subsequent to obtaining the duster set, reference documents are identified (block 42), The reference documents can include all reference documents generated for a document review project, or alternatively, a subset of the reference documents. Obtaining reference documents is further discussed below with reference to FIGURE 3.
An uncoded document is selected from one of the clusters in the set and compared against the reference documents (block 43) to identify one or more reference documents that are similar to the selected uncoded document (block. 44). The similar reference documents are identified based on a similarity measure calculated between the selected uncoded document and each reference document. Comparine the selected uncoded document with the reference documents is further discussed below with reference to FIGURE 4. Once identified, relationships between the selected uncoded document and the similar reference documents can be identified (block 45) to provide classification hints, including a suggestion for the selected ,uncoded document, as farther discussed below with reference to FIGURE 5.
Additionally, machine-generated suggestions for classification can be provided (block 46) with an associated.
confidence level for use in classifying the selected uncoded document. Machine-generated sugetestions are further discussed below with reference to FIGURE 7. Once the selected uncoded document is assigned a classification code, either by the reviewer or automatically, the newly classified document can be added to the set of reference documents for use in classifying further uncoded documents. Subsequently, a further uncoded document can be selected for classification using similar reference documents.
In a further embodiment, similar reference documents can also be identified for a selected cluster or a selected spine along Which the clusters are placed.
After the clusters have been generated, one or more uncoded documents can be selected from at least one of .the dusters for comparing with a .reference document set or subset.
FIGURE 3 is a block diagram Showing, by way of example, measures 50 for selecting a document reference subset 51. The subset of reference documents Si can be previously defined 54 and maintained for related document review projects or can be specifically generated for each review project. A predefined reference subset 54 provides knowledge previously obtained during the related document review project to increase efficiency, accuracy, and consistency.
Reference subsets newly generated for each review project can include arbitrary 52 or customized 53 reference subsets that are determined automatically or by a human reviewer. An arbitrary .reference subset 52 includes reference documents randomly selected for inclusion in the reference subset. A customized reference subset 53 includes reference documents specifically

9 selected for inclusion in the reference subset based on criteria, such as reviewer preference, classification category, document source, content, and review project. Other criteria are possible.
The subset of reference documents, whether predetermined or newly generated, should be selected from a set of reference documents that are representative of the document corpus for a review project in which data organization or classification is desired. Guided review assists a reviewer or other user in identifying reference documents that are representative of the corpus for use in classifying uncoded documents. During guided review, the uncoded documents that are dissimilar to all other uncoded documents are identified based on a similarity threshold. In one embodiment, the dissimilarity can be determined as the cos a of the score vectors for the uncoded documents. Other methods for determining dissimilarity are possible.
Identifying the dissimilar documents provides a group of documents that are representative of the corpus for a document review project. Each identified dissimilar document is then classified by assigning a particular classification code based on the content of the document to collectively generate the reference documents. Guided review can be performed by a reviewer, a machine, or a combination of the reviewer and machine.
Other methods for generating reference documents for a document review project using guided review are possible, including clustering. A set of uncoded documents to be classified is clustered, as described in commonly-assigned U.S. Patent No. 7,610,313. A
plurality of the clustered uncoded documents are selected based on selection criteria, such as cluster centers or sample clusters. The cluster centers can be used to identify uncoded documents in a cluster that are most similar or dissimilar to the cluster center. The selected uncoded documents are then assigned classification codes. In a further embodiment, sample clusters can be used to generate reference documents by selecting one or more sample clusters based on cluster relation criteria, such as size, content, similarity, or dissimilarity. The uncoded documents in the selected sample clusters are then selected for classification by assigning classification codes. The classified documents represent reference documents for the document review project. The number of reference documents can be determined automatically or by a reviewer. Other methods for selecting documents for use as reference documents are possible.
An uncoded document selected from one of the clusters can be compared to the reference documents to identify similar reference documents for use in providing suggestions regarding classification of the selected uncoded document. FIGURE 4 is a process flow diagram showing, by way of example, a method 60 for comparing an uncoded document to reference documents for use in the method of FIGURE 2. The uncoded document is selected from a cluster (block 61) and applied to the reference documents (block 62). The reference documents can include all reference documents for a document review project or a subset of the reference documents. Each of the reference documents and the selected uncoded document can be represented by a score 5 vector having paired values of tokens occurring within that document and associated token scores. A similarity between the uncoded document and each reference document is determined (block 63) as the cos a of the score vectors for the uncoded document and reference document being compared and is equivalent to the inner product between the score vectors. In the described embodiment, the cos a is calculated in accordance with the equation:
('A ' /3 )

10 cos cAB =
SA Sn where cos o-ġġ comprises a similarity between uncoded document A and reference document B, - A comprises a score vector for uncoded document A, and B comprises a score vector for reference document B. Other forms of determining similarity using a distance metric are possible, as would be recognized by one skilled in the art, including using Euclidean distance.
One or more of the reference documents that are most similar to the selected uncoded document, based on the similarity metric, are identified. The most similar reference documents can be identified by satisfying a predetermined threshold of similarity. Other methods for determining the similar reference documents are possible, such as setting a predetermined absolute number of the most similar reference documents. The classification codes of the identified similar reference documents can be used as suggestions for classifying the selected uncoded document, as further described below with reference to FIGURE 5. Once identified, the similar reference documents can be used to provide suggestions regarding classification of the selected uncoded document, as further described below with reference to FIGURES 5 and 7.
The similar reference documents can be displayed with the clusters of uncoded documents. In the display, the similar reference documents can be provided as a list, while the clusters can be can be organized along spines of thematically related clusters, as described in commonly-assigned U.S. Patent No. 7,271,804, the disclosure of which is incorporated by reference. The spines can be positioned in relation to other cluster spines based on a theme shared by those cluster spines, as described in commonly-assigned U.S. Patent No. 7,610,313.
Other displays of the clusters and similar reference documents are possible.

Organizing the clusters into spines and groups of cluster spines provides an individual reviewer with a display that presents the documents according to a theme while maximizing the number o.f relationships depicted between the documents. FIGURE 5 is a screenshot 70 showing, by way of exam.ple, a visual display 71 of similar reference documents 74 and uncoded documents 74. Clusters 72 of the uncoded documents 73 can be located along a spine, which is a vector, based on a .similarity of the uncoded documents 73 in the clusters 72.
The uncoded documents 73 are each represented by a smaller circle within the clusters 72.
Similar reference documents 74 identified for a selected uncoded document 73 can be displayed in a list 75 by document title or other identifier. Also, classification codes 76 associated with the similar reference documents 74 can be displayed as Circles havina a diamond Shape within the -boundary of the circle. The classification codes 76 can include "privileged,"
"responsive," and "non-responsive" codes, as well as other codes. The different classification codes 76 can each be represented by a color, such as blue for "privileged"
reference documents and ,yellow for "non-responsive" reference documents. Other display representations of the uncoded documents, similar reference documents, and classification codes are possible, including by symbols and shapes.
The classification codes 76 of the similar reference documents 74 can provide suestions for classifying .the selected uncoded document based on factors., such as a number of different classification codes for the similar reference documents and. a number of similar reference documents associated with each classification code. For example, the list of reference documents includes four similar reference documents identified for a particular uncoded document. Three of the reference documents are classified. as "privileged,"
while one is classified as "non-responsive." In making a decision to assign a classification code to a selected uncoded document, the reviewer can consider classification factors based on the similar reference documents, such as such as a presence or absence of similar reference documents with different classification codes and a quantity of the similar reference documents for each classification code. Other classification factors are possible. in the current example, the display 81 provides suggestions, including the number of "privileged" similar reference documents, the number of "non-responsive" similar reference documents, and the absence of other classification codes of similar .reference documents. Based on the number of "privileged"
similar reference documents compared to the number of "non-responsive" similar reference documents, the.
reviewer may be more inclined to classify the selected uncoded documents as "privileged."
Alternatively, the reviewer may wish to further review the selected uncoded document based on the multiple classification codes of the similar reference documents. Other classification codes and combinations of classification codes are possible. The reviewer can utilize the suggestions provided by the similar reference documents to assign a classification to the selected uncoded document In a .further embodiment, the now classified and previously uncoded document can be added to the set of reference documents for use in classifying other .uncoded documents.
In a further embodiment, similar reference documents can be identified for a chaster or spine to provide suggestions for classifying the cluster and spine. For a cluster, the similar reference documents are identified based on a comparison of a score vector for the cluster, which is representative of the cluster center and the reference document score vectors. Meanwhile, identifying similar reference documents for a spine is based on a comparison between the score vector for the spine, which is based on the cluster center of all tile clusters along that spine, and the reference document: score vectors. Once identified, the similar reference documents are used .for classifying the cluster or spine.
In an even further embodiment, the uncoded documents, including the selected uncoded document, and the similar reference documents can be displayed as a. document list. FIGURE 6 is a. screenshot 80 showing, by way of example, an alternative visual display of the similar reference documents 85 and um:Wed documents 82, The uncoded documents 82 can be provided as a list in an uncoded document box 81, such as an email inbox. The uncoded documents 82 can be identified and organized using uncoded document factors, such as file name, subject, date, recipient, sender, creator, and classification category 83, if previously assigned.
At least: one of the uncoded documents can be selected and displayed in a document viewing box 84. The selected uncoded document can be identified in the list 81 using a selection indicator (not Shown), including a symbol, font, or highlighting. Other selection indicators and uncoded document factors are .possible. Once identified, the selected uncoded document can be compared to a set: of -reference documents to identi.dy the reference documents 85 most similar.
The identified similar reference documents 85 can be displayed below the document viewing box 84 with an associated classification code 83. The classification code of the similar reference document 85 can be used as a suggestion for classifying the selected uncoded document. After assigning a classification code, a representation 83 of the classification can be provided in the display with the selected uncoded document, in a further embodiment, the now classified and previously uncoded document can be added to the set of reference documents.
Similar reference documents can be used as suggestions to indicate a need -for manual review of the uncoded documents, when review may be unnecessary, and hints for classifying the -uncoded documents, clusters, or spines. Additional information can be generated to assist a reviewer in making classification decisions for the uncoded documents, such as a machine-generated confidence level associated with a suggested classification code, as described in common-assigned U.S. Patent No. 8,635,223, entitled "System and Method for Providing a Classification Suggestion for Electronically Stored Information," filed on July 9, 2010.
The machine-generated suggestion for classification and associated confidence level can be determined by a classifier. FIGURE 7 is a process flow diagram 90 showing, by way of example, a method for classifying uncoded documents by a classifier for use in the method of FIGURE 2. An uncoded document is selected from a cluster (block 91) and compared to a neighborhood of x-similar reference documents (block 92) to identify those similar reference documents that are most relevant to the selected uncoded document. The selected uncoded document can be the same as the uncoded document selected for identifying similar reference documents or a different uncoded document. In a further embodiment, a machine-generated suggestion can be provided for a cluster or spine by selecting and comparing the cluster or spine to a neighborhood of x-reference documents for the cluster or spine.
The neighborhood of x-similar reference documents is determined separately for each selected uncoded document and can include one or more similar reference documents. During neighborhood generation, a value for x similar reference documents is first determined automatically or by an individual reviewer. The neighborhood of similar reference documents can include the reference documents, which were identified as similar reference documents according to the method of FIGURE 4, or reference documents located in one or more clusters, such as the same cluster as the selected uncoded document or in one or more files, such as an email file. Next, the x-number of similar reference documents nearest to the selected uncoded document are identified. Finally, the identified x-number of similar reference documents are provided as the neighborhood for the selected uncoded document. In a further embodiment, the x-number of similar reference documents are defined for each classification code, rather than across all classification codes. Once generated, the x-number of similar reference documents in the neighborhood and the selected uncoded document are analyzed by the classifier to provide a machine-generated classification suggestion for assigning a classification code (block 93). A
confidence level for the machine-generated classification suggestion is also provided (block 94).
The machine-generated analysis of the selected uncoded document and x-number of similar reference documents can be based on one or more routines performed by the classifier, such as a nearest neighbor (NN) classifier. The routines for determining a suggested classification code include a minimum distance classification measure, also known as closest neighbor, min iMUM average distance classification measures maximum count classification measure, and distance weighted maximum count classification measure. The minimum distance classification measure for a selected uncoded document includes identifying a neighbor that is the closest distance to the selected uncoded document and assigning the classification code of th.e closest neighbor as the suggested classification code for the selected uncoded document. The closest neighbor is determined .by comparing the score vectors for the selected uncoded document with each of the x-nurn.ber of similar reference documents in the .neighborhood as the cos a to determine a distance metric.. The distance metrics for the x-number of similar reference documents are compared to identify the similar reference document closest to the selected uncoded document as the closest .neighbor.
The minimum average distance classification measure includes calculating an average distance of the similar reference documents for each classification code. The classification code of the similar reference documents having the closest average distance to the selected uncoded document is assigned as .the suggested classification code. The maximum count classification measure, also known as the voting classification measure, includes counting a number of similar reference documents for each classification code and assigning a count or "vote" to the similar reference documents based on the assigned classification code. The classification code with the highest number of similar reference documents or "votes" is assigned to the selected uncoded document as the suggested classification code. Tile distance weighted maximum count classification measure includes identifying a count of all similar reference documents for each classification code and determining a. distance between the selected uncoded document and each of the similar reference documents. Each count assigned to the similar reference documents is weighted based on the distance of the similar reference document from the selected. uncoded document. The classification code with the highest count, after consideration of the weight, is assigned to the selected uncoded document as the suggested classification code.
The machine-generated suggested classification code is provided for the selected uncoded document.. with a confidence level, which can be presented as an absolute value or a percentage.
Other confidence level measures are possible. The reviewer can use the suggested classification code and confidence level to assign a classification to the selected uncoded document.
Alternatively, the x-NN classifier can automatically assign the suggested classification code. In one embodiment, the x-NN classifier only assigns an uncoded document with the suggested classification code if the confidence level is above a threshold value, which can be set by the reviewer or the x-NN classifier.

Machine classification can also occur on a cluster or spine level once one or more documents in the cluster have been classified. For instance, for cluster classification, a cluster is selected and a score vector for the center of the cluster is determined as described above with reference to FIGURE 4. A neighborhood for the selected cluster can be determined based on a 5 distance metric. The x-number of similar reference documents that are closest to the cluster center can be selected for inclusion in the neighborhood, as described above.
Each document in the selected cluster is associated with a score vector from which the cluster center score vector is generated. The distance is then determined by comparing the score vector of the cluster center with the score vector for each of the similar reference documents to determine an x-number of 10 similar reference documents that are closest to the cluster center.
However, other methods for generating a neighborhood are possible. Once determined, one of the classification routines is applied to the neighborhood to determine a suggested classification code and confidence level for the selected cluster. The neighborhood of x-number of reference documents is determined for a spine by comparing a spine score vector with the vector for each similar reference 15 document to identify the neighborhood of similar documents that are the most similar.
Providing classification suggestions and suggested classification codes has been described in relation to uncoded documents and reference documents. However, in a further embodiment, classification suggestions and suggested classification codes can be provided for the uncoded documents based on a particular token identified within the uncoded documents.
The token can include concepts, n-grams, raw terms, and entities. In one example, the uncoded tokens, which are extracted from uncoded documents, can be clustered. A token can be selected from one of the clusters and compared with reference tokens. Relationships between the uncoded token and similar reference tokens can be displayed to provide classification suggestions for the uncoded token. The uncoded documents can then be classified based on the classified tokens.
While the invention has been particularly shown and described as referenced to the embodiments thereof, those skilled in the art will understand that the foregoing and other changes in form and detail may be made therein.

Legal Events
DateCodeEventDescription
26 Jan 2012EEERExamination request