Text correlation calculation based on passage-level event representation

Zihao Zheng, Yang Li, Ming Liu, Bing Qin, Yitong LiuScientia 2020 · DOI

Abstract

With the explosion of web information, measuring text correlation is essential for filtering redundant content in information flow services. Traditional methods typically represent text as high-dimensional vectors, which often fail to capture the core event-based structure of news. This paper proposes a passage-level event representation method that constructs an event connection graph from sentence-level extractions. By identifying key nodes as representatives, the method accurately calculates text correlation. Experimental results show that this unsupervised approach outperforms conventional similarity measures and remains robust in fine-grained correlation tasks.

Introduction

In the era of massive information flows, filtering redundant news from multiple sources is a major challenge. Traditional methods typically represent text as high-dimensional vectors (like TF-IDF or word embeddings) and calculate similarity using cosine distance. However, news articles are fundamentally about events—who did what, to whom, and where. Standard vector models often fail to capture this structural "eventness." While recent research has moved toward event extraction, most work remains at the sentence level. To accurately measure if two articles are correlated, we need a way to represent the entire passage's event structure rather than just looking at isolated sentences or keyword frequencies.

Methodology

The proposed method uses a multi-stage pipeline to transform raw text into a structured event representation:

  1. Key Sentence Selection: Using an enhanced TextRank algorithm, the model identifies the most important sentences in a document to filter out noise like background information or boilerplate text.
  2. Sentence-Level Event Extraction: For each key sentence, the model extracts events using the ZORE model, identifying a trigger word (usually a verb) and its associated elements (entities/nouns).
  3. Event Polygon Construction: Each sentence-level event is represented as a "polygon" where the trigger word is the central node connected to its elements.
An event polygon representing the sentence-level relationship between a trigger (Visit) and entities (Trump, Britain, Wife).
  1. Passage-Level Graph Merging: The model merges these polygons into a Passage-Level Event Connection Graph. Connections are made if two events share the same entity or if their nodes have high semantic similarity (calculated via GloVe embeddings).
  2. Representative Selection: TextRank is applied to the final graph to rank nodes. The top-ranked trigger words, named entities, and nouns are selected as the "representatives" of the article.
  3. Correlation Calculation: To compare two articles, the model calculates the weighted average similarity between their respective sets of representatives.

Evaluation and Results

The researchers evaluated their approach on three datasets: Sogou News, ByteDance (Toutiao) News, and a manually annotated sports dataset designed to test fine-grained correlation.

  • Performance vs. Baselines: On the Sogou dataset, the model achieved an F1 score of 0.83, significantly higher than TF-IDF (0.67) and other keyword-based measures.
  • Robustness: While BERT-based models performed well on coarse categories, the graph-based approach was more robust on the manual dataset, where understanding specific "who-did-what" details was critical for determining correlation.
  • Unsupervised Advantage: Unlike deep learning models that require massive labeled datasets, this graph-based approach is unsupervised, making it highly adaptable to different domains without retraining.

Conclusion

By prioritizing structural "eventness" over word frequency, the proposed model captures the "semantic backbone" of a story while ignoring noisy words that don't contribute to the core narrative. Merging sentence-level events allows the model to stitch together the story's timeline and participants across the entire article. Future work will explore integrating deep learning embeddings directly into the graph construction process and applying the method to multi-document summarization.

Summary updated

Abstract

Along with the explosion of web information, information flow service has attracted the attention of users. In this kind of service, how to measure the correlation between texts and further filter the redundant information collected from multiple sources becomes the key solution to meet the users desire. Recently, the popular text correlation calculation methods mostly represent text as vector and then measure text similarity as text correlation. However, in information flow service, most of the texts are news, and the core element in a news is the event it stated. Therefore, we need a way to extract the core features that are related to the event stated by text, so we can accurately calculate text correlation via these extracted features. Unfortunately, recent event-related researches focus on the sentence-level. To calculate text correlation, we need to grasp the content of the text from the passage-level, which indicates that passage-level event analysis has more impact. To this end, we propose a passage-level event representation method based on sentence-level event extraction. It constructs a passage-level event connection graph based on the extracting results obtained from sentences. After that, it selects the important nodes in the graph as the representations of the passage-level events. Based on the passage-level representations, we can acquire text correlation. Experimental results indicate that our method outperforms conventional text correlation calculation methods.
BibTeX
@inproceedings{zheng2020text,
  author = {Zihao ZHENG and Yang Li and Ming Liu and Bing Qin and Yitong LIU},
  title = {{Text correlation calculation based on passage-level event representation}},
  booktitle = {Scientia},
  year = {2020},
  doi = {10.1360/ssi-2019-0272},
}
Figure