Document Recommendation Using Summarization Network and Ontology Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional recommendation systems fail to provide accurate recommendations from large and noisy source documents, lacking scalability and incorporation of background knowledge and discourse-level consistency.
Innovation Solution
A machine learning-based recommendation apparatus that generates a summary of a source document using a summarization network, filters candidate documents based on domain-specific knowledge and string similarity, and computes ranking scores for relevant document retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional recommendation systems process large and noisy source documents, then they can provide more comprehensive recommendations, but the accuracy and scalability deteriorate
Solution Approach 1:
The patent extracts and removes irrelevant words from the source document using a summarization network, keeping only the most important information. This extraction process eliminates noise while preserving essential content, thereby maintaining recommendation accuracy even when processing large documents.
Solution Approach 2:
The patent segments the source document into important and irrelevant parts through the summarization network, which identifies and separates key information from noise. This segmentation allows the system to process large documents efficiently by focusing computational resources on relevant segments.
2Reliability
If conventional recommendation systems incorporate background knowledge and discourse-level consistency, then recommendation quality improves, but system complexity increases
Solution Approach 1:
The patent introduces a filtering component as an intermediary between the summarization network and the recommendation engine. This intermediary incorporates domain-specific knowledge and discourse-level consistency information to filter candidate documents, improving recommendation quality without significantly increasing overall system complexity.
3Loss of information
If the system processes large and noisy source documents, then more information is available for recommendation, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary summarization and filtering of the source document before the main recommendation process. By pre-processing the document to extract and remove irrelevant information, the system reduces the computational burden of subsequent processing steps, thereby decreasing overall processing time while preserving essential information.
Data Source
AI summary
Systems and methods for item recommendation are described. One or more embodiments of the systems and methods include generating a hidden vector representation for each word of a source document; removing at least one word from the source document based on the hidden vector representation using a summarization network to obtain a summary document; filtering a plurality of candidate documents based on the source document to obtain a plurality of filtered candidate documents; comparing the summary document to each of the filtered candidate documents to obtain a ranking score for each of the filtered candidate documents; and identifying a relevant candidate document from the filtered candidate documents based on the ranking score.