Hypothesis Sentence Selection for Multi-Document Stance Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sentiment analysis techniques struggle to effectively form hypothesis sets from documents representing different stances, failing to capture nuanced differences and similarities in sentence stances across documents.
Innovation Solution
A method involving sentence clustering based on stance scores and similarity groups, selecting sentences with high stance variance and dissimilarity to form a hypothesis set that captures diverse document stances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sentences are clustered based on stance scores and similarity criteria, then the hypothesis set can capture diverse stances across documents, but the complexity of the clustering and selection process increases
Solution Approach 1:
The patent segments the document corpus into multiple clusters based on stance scores, then further divides each cluster into similarity groups. This segmentation allows the system to handle complex stance relationships by breaking down the entire document set into manageable, semantically coherent groups, thereby improving measurement precision while managing computational complexity through structured division.
Solution Approach 2:
The patent performs preliminary clustering of sentences into stance-based clusters before selecting hypotheses. By pre-organizing sentences according to their stance scores and similarity criteria, the system reduces the complexity of subsequent hypothesis selection and improves the accuracy of stance determination through systematic preparation of data structures.
2Adaptability or versatility
If sentences are selected based on high stance variance from clusters, then the hypothesis set represents diverse stances, but the computational requirements increase
Solution Approach 1:
The patent applies local quality by calculating stance variance specifically within each cluster and similarity group rather than across the entire document corpus. This localized approach allows the system to identify diverse stances within specific contextual groups, improving adaptability to different stance representations while reducing overall computational requirements by focusing calculations on relevant local structures.
Solution Approach 2:
The patent selects a subset of sentences with high stance variance from each similarity group to form the hypothesis set, rather than processing all sentences. This partial action approach maintains the representation of diverse stances while significantly reducing computational resources required, as only the most informative sentences need to be selected from each group.
Data Source
AI summary
Provided are a computer program product, system, and method for forming a hypothesis set from sentences across documents representative of different stances taken across the documents. Sentences from the documents are clustered into a plurality of clusters. Sentences in a cluster of the clusters have stance scores with respect to other sentences in the cluster that satisfy a stance criteria. At least one similarity group of sentences is formed in the clusters having similarity scores satisfying a similarity criteria. Sentences are selected from the similarity groups in the clusters based on stance scores of the sentences in a similarity group. A hypothesis set is formed of the selected sentences in the similarity groups. Stance scores are determined of sentences in the documents with the sentences in the hypothesis set to determine stances of the documents with respect to the sentences in the hypothesis set.