Email Thread Hierarchy Detection via Segment Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing email management systems lack an efficient method to identify and organize email threads across a collection of emails, particularly when emails are forwarded or replied to with minimal edits, making it difficult to find similar emails.
Innovation Solution
A method and system that parse email content to generate segments, create segment signatures, and build an email thread hierarchy by comparing query segments with a pre-defined token match threshold, using a signature index to identify and organize related emails.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If email content is parsed and segmented to create signatures for thread hierarchy detection, then email retrieval efficiency is improved, but system complexity increases
Solution Approach 1:
The patent divides email content into multiple segments (e.g., subject line, body, attachments) and creates signatures for each segment. This segmentation allows the system to efficiently compare specific parts of emails rather than processing entire emails, improving retrieval efficiency while managing complexity through modular processing
Solution Approach 2:
The patent introduces segment signatures as intermediary representations that mediate between raw email content and the thread hierarchy detection process. These signatures serve as compact, comparable descriptors that reduce the complexity of direct email content comparison while maintaining the ability to detect thread relationships
2Measurement precision
If segment signatures are created and stored in a signature index, then email clustering accuracy is improved, but memory usage increases
Solution Approach 1:
The patent extracts essential characteristics from email segments to create compact signatures that capture the core identity of each segment. By taking out only the essential features (key tokens, segment type identifiers) rather than storing complete segment content, the system achieves accurate clustering while minimizing memory consumption
Solution Approach 2:
The patent transforms email segment data into a different parameter space by creating signatures with specific properties (token frequency, segment type codes, hierarchical identifiers). This parameter transformation enables precise clustering comparisons while representing email content in a memory-efficient format
3Measurement precision
If token match threshold comparison is used to identify root segments, then thread detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent applies a token match threshold that requires only partial matching of tokens between segment signatures rather than complete matching. This partial action approach achieves sufficient thread detection accuracy by identifying key common tokens while avoiding the excessive processing time that would result from requiring complete token sequence matching
Data Source
AI summary
A plurality of segments in an e-mail collection by parsing content of e-mails is generated. Corresponding segment signature for each segment is created and a signature index is populated using the generated segment signatures. After receiving a query e-mail, a plurality of query segments in the query e-mail is generated using content of the query e-mail and corresponding query segment signature for each query segment is generated. A query root segment is identified and corresponding query root segment signature is generated. A set of root segment signatures of the signature index is identified and the query root segment signature is compared with each root segment signature from the signature index. A subset of the signature index is identified, using a match between the root segment signature and the query root segment signature. An e-mail thread hierarchy is built using the identified subset of the signature index.