Overview
DISCO AI indexes document content and groups similar topics into clusters. The interface displays these clusters as a list, serving as a table of contents. Users browse broad topics or search specific ones to find documents. Each cluster appears as a row, reducing review time and helping locate relevant files.
- Explore topic clusters during early case assessment to identify custodians and date ranges, guiding the review strategy.
- Identify relevant clusters to prioritize workflows and exclude irrelevant ones.
- Search clusters to locate key documents during evidence gathering.
Identifying topics in documents
DISCO identifies topic clusters during document ingestion without user input. Clustering starts with 20,000 eligible review documents. The system excludes certain file types, like images and numeric spreadsheets. Topic clustering analyzes topics at the sentence level, allowing multiple topics per document. Documents can belong to multiple clusters. Tagging decisions affect AI tag predictions but do not change clusters.
Searching with topic clusters
Search visualizations and filters display topics. Combine topics with date ranges, document text, and other criteria. A topic cluster search returns documents matching that content.
The topic list organizes clusters from broad to specific. Broad clusters contain sub-topics. Expand parent clusters to view child clusters.
The histogram bar displays the document count for that topic cluster. The percentage value quantifies the topic cluster document count against the total search results.
Navigating topic clusters
Enter keywords to locate matching topic clusters. The system filters topic clusters in this example to those containing the word market. DISCO maintains the topic hierarchy to support continued browsing.
Maintaining topic clusters
Add or remove documents from the corpus. DISCO evaluates new documents against existing topic clusters. The system assigns matching documents to these clusters. Filter documents by ingestion batch to review assigned topics.
Corpus modifications trigger the generation of a new topic list. DISCO AI assesses the corpus to schedule topic list updates. The system displays the current topic list and an update notification during generation. Users preserve document sets using folders. The system activates the new topic list and deletes the previous version.