Document Clustering using a Graph Covering with Pseudostable Sets

Abstract

Abstract-In text mining, document clustering describes the efforts to assign unstructured documents to clusters, which in turn usually refer to topics. Clustering is widely used in science for data retrieval and organisation. In this paper we present a new graph theoretical approach to document clustering and its application on a real-world data set. We will show that the wellknown graph partition to stable sets or cliques can be generalized to pseudostable sets or pseudocliques. This allows to make a soft clustering as well as a hard clustering. We will present an integer linear programming and a greedy approach for this NP-complete problem and discuss some results on random instances and some real world data for different similarity measures.

Docclustering: An Implementation of the Novel Ps-Document Clustering Algorithm a Preprint

“…The Integer Linear Program Approach, introduced in [1] and implemented in DocClustering, can calculate optimal solutions of the multiple pseudostable sets partitioning problem (minMPS'-a-IP as given in [1]). It defines constraints to all relevant steps of the underlying graph partition and has exponential runtime.…”

“…It is used to assign documents to clusters, which usually represent topics. In [1] a novel approach using graph theory to document clustering is presented by Dörpinghaus et al and discussed together with its application on a real-world data set retrieved from the PubMed database. It is shown that this method is superior to conventional algorithms and provides more accurate clustering on biological and medical data, depending on the chosen similarity measure.…”

“…Then, the technical clustering process can be performed using an approach of graph theory. Jain et al also suggest a final step called "assessment of output" and it is shown that this can also be solved using graph theory and building the graph visualization proposed in [1].…”

“…Hirsch et al, cover related problems like clustering in the context of search queries [6], whereas Lee et al work on the field of hierarchical clusterings [7]. Thus, it can be concluded that focusing only on graph clustering is a novel approach and the generalization of soft document clustering introduced in [1] provides new insights into document clustering -or clustering in general -from the developed graph theoretical toolbox .…”

“…In this section, the background of the algorithms used to solve the clustering process is briefly described. For a more detailed overview of the approaches please see [1]. The reading process reads all relevant data and computes the document graph G where each node represents a document.…”