Content Relevance Models Using Context Beyond Exact Matches
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search engines struggle to provide a comprehensive overview of documents surrounding a particular topic due to their reliance on exact word matches, often missing relevant documents.
Innovation Solution
A content relevance model is developed using key word sets and context-based word pairs to determine the relevance of content segments to a specific category, involving the identification of anchor words and context definitions to score document relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search engines use exact word matches to find documents, then search results are precise for exact keyword queries, but relevant documents that use different wording are completely overlooked
Solution Approach 1:
The patent changes the parameter of document representation from exact word matching to semantic feature extraction. By representing documents as vectors of semantic features (concepts, entities, relationships) rather than just word tokens, the system can retrieve documents that are semantically relevant even when they use different wording, thus improving versatility while maintaining precision through semantic similarity metrics.
Solution Approach 2:
The patent introduces semantic features as an intermediary between user queries and documents. Instead of directly matching words to words, the system translates both queries and documents into semantic feature representations, then compares these intermediaries to find relevant documents. This intermediary layer enables the system to understand semantic relationships and retrieve relevant documents that use different wording.
2Productivity
If search engines rely on word and operator matching, then the search system remains simple and fast, but it cannot provide a good overview of documents surrounding a particular topic
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing semantic feature representations for documents during an indexing phase. This preliminary processing allows the system to quickly compare semantic features during actual search operations without needing to perform complex analysis in real-time. The semantic features are extracted and stored beforehand, enabling fast retrieval while providing comprehensive topic overviews through semantic similarity search.
3Ease of operation
If search engines use only words and operators entered by users, then the search interface remains simple, but many relevant documents are completely overlooked
Solution Approach 1:
The patent implements self-service by having the search system automatically analyze and interpret the semantic meaning of user queries without requiring explicit guidance. The system autonomously extracts semantic features from queries and compares them with document representations to find relevant documents, even when users don't provide precise keywords. This maintains interface simplicity while significantly improving reliability through automated semantic understanding.
Data Source
AI summary
Some embodiments provide a method for defining a content relevance model for determining whether a content segment is relevant to a particular category. The method receives a first set of content segments that contain content relevant to the particular category and a second set of content segments that contain content not relevant to the particular category. The method identifies a set of key word sets more likely to appear in the first set of content segments than the second set of content segments. The method defines a content relevance model that comprises a set of groups of word sets and a score for each group, each of the groups of word sets comprising a key word set from the set of key word sets and at least one word set found in a context of the key word set in at least one of the received content segments.