[Paper Review] PDC -- a probabilistic distributional clustering algorithm: a case study on suicide articles in PubMed

Rezarta Islamaj, Lana Yeganova+4 more|arXiv (Cornell University)|Dec 4, 2019

Biomedical Text Mining and Ontologies11 references4 citations

TL;DR

PDC is a probabilistic distributional clustering algorithm that groups terms from biomedical texts based on word co-occurrence probabilities, enabling topic discovery in large document collections. Applied to suicide-related PubMed articles, it identifies coherent topic clusters and supports retrieval of relevant literature through an interactive web portal, offering researchers a structured view of mental health literature.

ABSTRACT

The need to organize a large collection in a manner that facilitates human comprehension is crucial given the ever-increasing volumes of information. In this work, we present PDC (probabilistic distributional clustering), a novel algorithm that, given a document collection, computes disjoint term sets representing topics in the collection. The algorithm relies on probabilities of word co-occurrences to partition the set of terms appearing in the collection of documents into disjoint groups of related terms. In this work, we also present an environment to visualize the computed topics in the term space and retrieve the most related PubMed articles for each group of terms. We illustrate the algorithm by applying it to PubMed documents on the topic of suicide. Suicide is a major public health problem identified as the tenth leading cause of death in the US. In this application, our goal is to provide a global view of the mental health literature pertaining to the subject of suicide, and through this, to help create a rich environment of multifaceted data to guide health care researchers in their endeavor to better understand the breadth, depth and scope of the problem. We demonstrate the usefulness of the proposed algorithm by providing a web portal that allows mental health researchers to peruse the suicide-related literature in PubMed.

Motivation & Objective

• To address the challenge of organizing vast biomedical literature collections for improved human comprehension.
• To develop a method that automatically identifies topic-relevant term clusters using word co-occurrence probabilities.
• To create a visual and interactive environment for exploring suicide-related mental health literature in PubMed.
• To support researchers in gaining a comprehensive, multifaceted understanding of the scope and depth of suicide-related research.
• To demonstrate the utility of the algorithm through a case study on suicide-related articles in PubMed.

Proposed method

• The algorithm computes term co-occurrence probabilities across a document collection to model semantic relationships between words.
• It uses a probabilistic framework to partition the set of unique terms into disjoint, topic-representative groups.
• The method relies on distributional semantics, where terms with similar co-occurrence patterns are grouped together.
• A visualization environment is built to represent the term clusters in a 2D space, enabling intuitive exploration.
• The system retrieves and displays the most relevant PubMed articles for each cluster based on term similarity.
• An algorithmic appendix details the computational steps, including probability estimation and clustering logic.

Experimental results

Research questions

• RQ1 Can a probabilistic distributional clustering approach effectively group terms into coherent topics within a large biomedical text collection?
• RQ2 How well does the PDC algorithm identify meaningful topic clusters in suicide-related mental health literature?
• RQ3 To what extent does the resulting topic structure support human comprehension and literature exploration in mental health research?
• RQ4 Can the system provide an intuitive, interactive interface for retrieving and browsing relevant PubMed articles by topic?
• RQ5 How does the PDC approach compare to traditional clustering methods in organizing biomedical text data?

Key findings

• The PDC algorithm successfully partitions terms from suicide-related PubMed articles into distinct, interpretable clusters reflecting key themes in mental health research.
• The method identifies coherent topic groups such as risk factors, prevention strategies, and treatment modalities in suicide literature.
• The web portal enables researchers to explore clusters and retrieve relevant articles, enhancing access to multifaceted data.
• The visualization of term clusters in a 2D space provides an intuitive, scalable way to navigate complex literature.
• The system demonstrates the feasibility of using probabilistic distributional clustering for organizing and exploring large-scale biomedical literature.
• The case study confirms that the algorithm supports a global, structured view of a complex research domain like suicide-related mental health.