Electronic Discovery - Review

Using MetaLINCS search and analysis tools, a case manager or attorney can quickly determine key people, terms and documents. This case assessment is used to develop a review strategy and organize the documents based on that strategy. Typical review categories that can quickly be determined include: topic clusters, possible privilege, hot, possibly responsive, low priority and specific foreign languages. Within each of these broad categories specific folders can be created to further divide the document sets for individual reviewers. The ability to create custom review folders and bulk move any number of items that match a custom criteria into those folders makes this organization phase very efficient.

An unlimited set of custom tags, can be defined per case. Flags can be grouped in exclusive or non-exclusive groups. Exclusive groups only allow a reviewer to pick one flag from the group while non-exclusive groups allow multiple flags. A typical exclusive group would include Privilege, Responsive, Non-Responsive.

Most items in a review are emails and most emails are part of a thread. Tagging by thread allows a reviewer to mark all the messages, and by default attachments, in a thread with a single operation. Reviewers can understand individual emails in the context of an entire thread and review by thread, increasing review speed and accuracy.  A graphical and text oriented presentation of the information shows both the structure and content of a thread.

Reviewers can share information via bookmarks which can save a reference to a search request, specific document, thread or person. Bookmarks can be organized in hierarchical folders that other reviewers can open. This allows complex searches to be saved and shared among reviewers as well making the results of those searches easy to retrieve.

MetaLINCS supports Unicode and automatic language detection. MetaLINCS converts all documents to Unicode for internal storage. Any language can be indexed for search and its' character set can be displayed. Stemming is supported for many languages, including: English, French, Italian, German, Spanish, Danish, Dutch, Norwegian, Portuguese, so that multiple forms of the same word will be found, e.g. run, running, runs. Documents and messages written in these languages are automatically recognized so that the correct stemmer is applied. Automatic detection also means that searches can be done for all documents written in a specific language. Japanese, Chinese and Korean are all automatically recognized and can be indexed, displayed and searched. Japanese and Chinese are accurately indexed even though they generally do not have spaces between words.

MetaLINCS Analysis is at the heart of a winning strategy:

Sophisticated search, analysis, and categorization capabilities, including Guided Navigation and analysis based on:
      - Concepts and Concept Clusters
      - Email Threads
      - Social Network Analysis
      - Extracted Entities

Rapid identification of key case topics, people, vocabulary, and documents along with automated culling of irrelevant information.
Immediate access to email analysis and organized case content.
Flexible case organization to provide context during the review process.
Intuitive navigation and a user-interface based on familiar tools.

MetaLINCS combines a unique search and browse interface with extensive search algorithms to quickly find emails and other documents. MetaLINCS embraces all of the traditional search methods including keywords, phrases, concepts, Boolean combinations, near queries, range queries, wildcard queries, spelling-fuzzy queries and then goes far beyond traditional search to show the relationships between items and people, and between related items. This advanced search capability allows users to scan the search results by topics, threads, people and concepts. Dynamically generated high-level views of the search results help a user develop a glossary of key terms while determining if a set of search results requires further scrutiny. All 250-odd content and metadata fields that MetaLINCS defines for items and people are searchable. Values in these fields can be identified by a range of query operators. There are specialized types of queries for identifying desired items by how they relate to other items. MetaLINCS has patents pending on its techniques for querying items based on their relationships.

MetaLINCS Guided Navigation helps reviewers and investigators find key information in lightning speed and in any direction within the directory.  With each new selection, the MetaLINCS directory updates to provide a dynamic picture of the resulting document set. This constantly updating overview of the document set provides guidance on where the most important documents are located and what they contain. The dynamic nature of the directory is in contrast to traditional static directories where selecting any particular category reduces not just the result set but also the set of potentially useful ways to categorize that result set. Selectors are also ranked with a numeric "score" indicating the prevalence of each selector.

People are frequently as important an object type as content items. MetaLINCS provides a number of analysis tools that take advantage of its people object model to help users identify the key people associated with the matter under discovery, the roles these people played, and the time frames and specifics of their involvement. The Directory identifies key people with statistics about their information role, both for the collection as a whole and for any sub-collection identified by search and/or drill-down. For example, a query can be expressed to retrieve all items relevant to a particular issue and MetaLINCS will find both the items and the associated people. With Social Network Analysis, users can see patterns in communication between and among custodians and the results can be graphed to show the changing patterns over time.

Thread Analysis
Email threads are more natural units to understand what happened than single messages. They provide the context surrounding each individual message and attached document, illuminating the events and roles of the people involved. Items important to a case are frequently not the same as items that have high search relevance. Important items may not use explicit terminology and are frequently private communications that involve only a small number of participants. However, these items are frequently offshoots of larger public conversations, i.e. “side conversations” within a broader context. MetaLINCS Thread AnalysisTM organizes all of the messages in a thread, including all the branches in the thread that were created when messages were forwarded. Viewing the messages by thread lets the user quickly determine who all of the participants were, what attachments they received and graphically see the temporal relationship between the various related discussions. Displaying the text of messages alongside the graphical presentation accelerates the process of understanding the conversation and the context around each message in the thread.

Hierarchy Document Clustering and Review
Multi-lingual topical document clusters provide an alternative to traditional custodian-based review. Grouping documents by topic helps reviewers make faster better decisions on how to flag a document because they see the document in the context of many similar documents. The document clusters are hierarchical so subsets within a cluster can be assigned to individual reviewers. A special UI available to the case manager displays the topic hierarchy and a relevant label is automatically created for each cluster. Document clusters are formed on a streaming basis as documents are processed, rather than on a batch basis after processing has finished. This allows analysis and early case assessment to start sooner. Clustering is integrated with search and supports two review workflows. Reviewers can select a custodian folder and review by topic within that folder or bypass custodian folders and review purely by global topic.