Hyperlink Vectorization for Citation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for processing hypertext documents result in information loss by ignoring the relationship between hyperlink content and its source and context, leading to incomplete representation and intention loss in citation recommendation and retrieval tasks.
Innovation Solution
A hyperlink processing method that converts context information of hyperlinks into context vectors, adjusts input and output vectors, and calculates similarity to ensure accurate representation and retention of information, using a combination of vectorization techniques like h-d2v to model source, context, and target documents directly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If hyperlink documents are converted into plain objects for vectorization, then the processing simplicity is improved, but information loss occurs
Solution Approach 1:
The patent segments the hyperlink document into multiple components: the original content, the hyperlink object, the source document information, and the target document information. Each component is vectorized separately and then combined to form a comprehensive representation that preserves all hyperlink-related information while maintaining processing simplicity.
2Device complexity
If hyperlink relationships are ignored in vectorization, then the processing complexity is reduced, but citation recommendation accuracy deteriorates
Solution Approach 1:
The patent merges multiple vector representations (content vector, source document vector, target document vector, and hyperlink object vector) into a unified comprehensive vector representation. This combining approach captures the full hyperlink relationship context without significantly increasing processing complexity, thereby improving citation recommendation accuracy.
3Productivity
If context information of hyperlinks is discarded, then the computational efficiency is improved, but context intention sensitivity deteriorates
Solution Approach 1:
The patent extracts the context information surrounding hyperlinks (such as the sentences or paragraphs containing the hyperlinks) and incorporates it into the vector representation. This extraction process captures the contextual intent without significantly impacting computational efficiency, as the extracted context is processed using the same efficient vectorization pipeline.
Data Source
AI summary
Embodiments of this application disclose a hyperlink processing method and apparatus. The method includes: converting first context information of a first hyperlink in a first object into a first context vector; obtaining a first input vector of the first object, the first object including information on the first hyperlink pointing to a second object; obtaining a first average vector according to the first context vector and the first input vector; adjusting at least one of the first input vector, the first context vector, or a first output vector corresponding to the second object; and calculating a similarity between the first output vector and the first average vector according to an adjustment result, and outputting the first output vector as an output vector of the second object in response to the similarity being greater than or equal to a first target threshold.