Thread Information
The Thread Information Panel includes Thread details, Thread diagram, Near duplicate documents and Similar threads, as well as Comments, Tags, Entities and Precision Hits.
1. Threads Details Window
The Thread details window has the following field columns:
Hit: if the document is a direct search hit.
Inclusive: if the document is an inclusive email, that is, not missing any attachments or content of items in the stage of the thread that it represents.
Control Number: the unique internal identifier of the document.
Subject: the “subject“ line of email.
Sent Date: the “Date Sent” property of the email.
Cluster: ID of the cluster.
Cluster Score: score of clustering, high score indicates it is closer to the center of the cluster.
Attach Count: number of attachments.
Seg Count: number of segments in the email.
Model Score: the currently selected COSMIC model score.
Cycle: number of cycle for the current COSMIC project.
Group: group identifier.
Attachment: if this is an attachment or not.
Use the button to show or hide columns in the Thread details window:
2. Thread Diagram
The Thread Diagram is located below the Thread details window.
Each circle in the diagram represents a segment from the document.
Each unique path along the connecting lines from the furthest left circle in the top row to the furthest right circle in each row represents a unique document. In addition, an individual row may contain more than one document as indicated.
The Blue dots on the top indicate that the segment(s) are search hits.
The yellow circle is the currently loaded segment.
For example, the visualization above indicates the thread has 3 documents. Document A has segments 1, 2, 3 and 4, document B has segment 1, 5 and 6, and document C has segment 1, 2, 7.
Select any of the circles to load the selected segment into the document viewer. Even if you select a segment that is not a part of the original thread such as Segment 5 in the above visualization, it will show you that segment.
3. Near Duplicate Documents
Documents which are substantially similar to the current one may be examined from the Near duplicate documents pane. The table lists the documents by Control Number and Similarity Score, Under Similarity Score, the most complete Anchor document is assigned a value of "1", and the remainder are listed in descending order of similarity to the anchor.
In general, Near duplicates with a similarity score of .8 or above are deemed to be significant. These may be drafts of a document or variations on a message worth noting for their changes. To directly examine the near duplicates in a thread, click on the View all <#> button. The viewer displays a banner indicating that the near duplicate thread is active.
4. Similar Threads
Click the Similar threads button to request threads related to the currently loaded one based on when this document was sent, its communicators, the entities detected, or the cluster it's in.
The similar threads found will be loaded into the documents snippet window to the right:
For information on Tags, see Reviewing & Tagging Documents in COSMIC. For Entities, see User Annotation.