A semantic similarity approach to predicting Library of Congress subject headings for social tags

Abstract

Social tagging or collaborative tagging has become a new trend in the organization, management, and discovery of digital information. The rapid growth of shared information mostly controlled by social tags poses a new challenge for social tag-based information organization and retrieval. A plausible approach for this challenge is linking social tags to a controlled vocabulary. As an introductory step for this approach, this study investigates ways of predicting relevant subject headings for resources from social tags assigned to the resources. The prediction of subject headings was measured by five different similarity measures: tf-idf, cosine-based similarity (CoS), Jaccard similarity (or Jaccard coefficient; JS), Mutual information (MI), and information radius (IRad). Their results were compared to those by professionals. The results show that a CoS measure based on top five social tags was most effective. Inclusions of more social tags only aggravate the performance. The performance of JS is comparable to the performance of CoS while tf-idf is comparable with up to 70% less than the best performance. MI and IRad have inferior performance compared to the other methods. This study demonstrates the application of the similarity measuring techniques to the prediction of correct Library of Congress subject headings.

Dealing with the long tail: Providing uniformity to compound tags

“…With this point of view, studies have compared user tags with controlled vocabularies, such as subject headings (Lee & Schleyer, 2010;Lu, Park, & Hu, 2010;Rolla, 2009;Yi & Chan, 2009), using different measurements for comparison including Jaccard similarity, TF-IDF, cosine-based similarity, etc. (Lee & Schleyer, 2010;Razikin et al, 2011;Yi, 2011).…”

Making sense of digital traces: An activity theory driven ontological approach

“…Emergent research on the use of semantics to make sense of digital traces has linked social tags to a controlled vocabulary (Yi, 2010) and uses tag analysis to uncover semantic relations (Yoon, 2012).…”

Conceptual syntagmatic associations in user tagging

“…Many studies on tagging were interested in integrating tagging into the development of an ontology for information systems in the context of knowledge organization (e.g., Tsui et al., ; Yi, ) [“Ontology” is used to mean a vocabulary structure of concepts and their relations for an information system, which is more precisely a thesaurus and taxonomy (Fonseca, ), hence, folksonomy, i.e., taxonomy of people.] Many researchers in this line have pointed out folksonomy as an alternative to augment, if not replace, traditional taxonomies and hierarchical classification of content description because it better addresses issues of scalability in a rapidly changing Internet environment (Ding et al., , ; Tonkin, Corrado, & Moulaison, ; Yi, ).…”

“…Many studies on tagging were interested in integrating tagging into the development of an ontology for information systems in the context of knowledge organization (e.g., Tsui et al., ; Yi, ) [“Ontology” is used to mean a vocabulary structure of concepts and their relations for an information system, which is more precisely a thesaurus and taxonomy (Fonseca, ), hence, folksonomy, i.e., taxonomy of people.] Many researchers in this line have pointed out folksonomy as an alternative to augment, if not replace, traditional taxonomies and hierarchical classification of content description because it better addresses issues of scalability in a rapidly changing Internet environment (Ding et al., , ; Tonkin, Corrado, & Moulaison, ; Yi, ). These studies have considered tags as comprising a flat list of terms in which no explicit relations among the terms were evident; however, as Peters and Weller () stated, “…still, we can assume that a user does keep the idea of certain relationships in mind during the process of tagging content.” The current study addresses the conceptual associations among the terms and the meanings behind the terms that might implicitly represent the user meaning attached to the information object within the term relations.…”

“…Taken together as a group, terms may indicate a semantic relationship and associations among themselves, which would be useful for content description even though the co‐occurrence of terms does not necessarily identify an explicit relationship (Bruns, ). One recent study (Yi, ) has taken semantic relations among tag terms into consideration to integrate social tags into the existing controlled vocabulary; it took into account the pre‐existing relations (e.g., equivalence and hierarchical relationships), but did not include associative verb relations. In this study, the focus was on such associations that are not specifically predefined but syntagmatically observable, which will be described in the following section.…”