No abstract

Document clustering addresses the problem of identifying groups of similar documents without human supervision. Unlike most existing solutions that perform document clustering based on keywords matching, we propose an algorithm that considers the meaning of the terms in the documents. For example, a document that contains the words "dog" and "cat" multiple times may be placed in the same category as a document that contains the word "pet" even if the two documents share only noise words in common. Our semantic clustering algorithm is based on a similarity graph that stores the degree of semantic relationship between terms (extracted from WordNet), where a term can be a word or a phrase. We experimentally validate our algorithm on the Reuters-21578 benchmark, which contains 11, 362 newswire stories that are grouped in 82 categories using human judgment. We apply the k-means clustering algorithm to group the documents using a similarity metric that is based on keywords matching and one that uses the similarity graph. We show that the second approach produces higher precision and recall, which means that this approach matches closer the results of the human study.

No abstract

The paper addresses the problem of modeling the relationship between phrases in English using a similarity graph. The mathematical model stores data about the strength of the re lationship between phrases expressed as a decimal number. Both structured data from Wikipedia, such as that the Wikipedia page with title "Dog" belongs to the Wikipedia category "Domesticated animals", and textual descriptions, such as that the Wikipedia page with title "Dog" contains the word "wolf" thirty one times are used in creating the graph. The quality of the graph data is validated by comparing the similarity of pairs of phrases using our software that uses the graph with results of studies that were performed with human subjects. To the best of our knowledge, our software produces better correlation with the results of both the Miller and Charles study and the WordSimilarity-353 study than any other published research.

We propose a novel way for extracting the strength of the semantic relationship between words from semi-structured sources, such as WordNet. Unlike existing approaches that only explore the structured information (e.g., the hypernym relationship in WordNet), we present a framework that allows us to utilize all available information, including natural text descriptions. Our approach constructs a similarity graph that stores the strength of the semantic relationship between words. Specifically, an edge between two words describes the probability that someone who is interested in resources about the first word will be also interested in resources about the second word. Note that the graph is asymmetric because the probability that someone is interested in the second word given that they are interested int the first word is not the same as the probability that they are interested in the first word given that they are interested in the second word. The similarity between any two words in the graph can be computed as a function of the directed paths between the two nodes in the graph that represent the words.We evaluate the quality of the data in the similarity graph by comparing the similarity of pairs of words using our software that uses the graph with results of studies that are performed with human subjects. To the best of our knowledge, our software produces better correlation with the results of both the Miller and Charles study and the WordSimilarity-353 study than any other published research. We also present an extended evaluation section that describes how the different heuristics that we use affect the correlation score.