Sentence Importance Evaluation Using Modified PageRank

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional document summarization methods fail to account for reader preferences, leading to the extraction of important sentences that may not be relevant to individual readers, resulting in inconsistent and inaccurate summaries.

Innovation Solution

A method and system that calculate the importance of each sentence in a document using a modified PageRank algorithm, considering the relevance of adjacent sentences and keywords, to extract and summarize sentences based on reader-specific preferences, implemented in a document summarization apparatus with network interface, processors, and memory.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a general extraction algorithm is used to extract important sentences from a document, then sentences evaluated as important by absolute standards can be extracted, but the extracted sentences may not be relevant to individual readers' interests and intentions

Engineering Contradiction:

Improveaccuracy of sentence importance evaluationVSAvoidadaptability to reader preferences

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent applies dynamics by making the sentence extraction process adaptable and flexible rather than static. The system dynamically adjusts the importance evaluation of sentences based on reader profiles, interests, and intentions. The graph structure and PageRank algorithm are enhanced to incorporate reader-specific parameters, allowing the extraction results to change dynamically according to different reader characteristics.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent changes parameters by introducing reader-specific parameters (interests, intentions, preferences) into the sentence importance evaluation process. The modified PageRank algorithm incorporates these parameters as weighting factors, changing the evaluation criteria from absolute standards to relative standards that adapt to different readers. This allows the same document to yield different important sentences for different readers.

Inventive Principle:

Principle #35Parameter changes

2Stability of the object's composition

If extraction methods are used to generate summaries, then the summaries maintain consistency with the original document, but they may not accurately represent the information most relevant to specific readers

Engineering Contradiction:

Improveconsistency with original documentVSAvoidrelevance information loss for specific readers

Core Design Contradiction:

Stability of the object's compositionVSLoss of information

Solution Approach 1:

The patent applies local quality by differentiating the treatment of sentences based on their relevance to specific readers. Instead of applying a uniform extraction criterion to all sentences, the system evaluates each sentence's importance relative to the reader's profile, interests, and intentions. This allows the summary to maintain consistency with the document while selectively emphasizing sentences that are most relevant to each reader.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent incorporates feedback mechanisms by using reader profiles and preferences to guide the sentence extraction process. The system takes feedback from reader characteristics and adjusts the extraction criteria accordingly. The modified PageRank algorithm uses reader-specific parameters as feedback to refine the importance evaluation, ensuring that the extracted sentences align with reader expectations and needs.

Inventive Principle:

Principle #23Feedback

3Adaptability or versatility

If a modified PageRank algorithm incorporating reader preferences is used, then personalized summaries can be generated, but the calculation complexity and processing time increase

Engineering Contradiction:

Improvepersonalization capabilityVSAvoidalgorithm complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-processing and structuring the document into a graph representation before applying the modified PageRank algorithm. The graph structure with vertices and edges is established in advance, and reader profiles are pre-analyzed to extract relevant parameters. This preliminary preparation reduces the computational burden during the actual extraction process, as the complex graph structure and reader parameters are already organized and ready for efficient processing.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10430468B2Method and system for extracting sentences

Publication Date: 2019.10.01 UBERPLE

AI summary

Methods and systems for extracting sentences are provided, one of methods comprises, receiving a keyword, parsing a document, and identifying each of a plurality of sentences included in the parsed document, configuring a graph having vertices and edges, wherein each vertex corresponds to each sentence, and each edge has a first weight corresponding to similarity between each pair of the sentences, calculating importance of each sentence by applying a modified PageRank algorithm to the graph, wherein the modified PageRank algorithm is designed to reflect a second weight corresponding to whether the keyword is included in a sentence of each vertex adjacent to a first vertex and extracting important sentences from the document based on the calculated importance.