Learning Concept Hierarchy from Folksonomy

Abstract

Users often use tags to annotate and categorize web content. A folksonomy is a system of classification derived from the practice and method of collaboratively creating and managing tags. The most significant feature of a folksonomy is that it directly reflects the vocabulary of users. This feature is very useful in tag-based content searching and user browsing. Based on mutual-overlapping measurement of tag's instance sets, an ontology learning algorithm to construct concept hierarchy from folksonomy is proposed. A case study of datasets from a famous Chinese e-business website taobao is carried out. The precision, valid, recall and F-measure rates of the constructed concept hierarchy are 54%, 84%, 100% and 70% respectively. The experimental results on real world datasets show that the proposed method is feasible.

A supervised learning approach to detect subsumption relations between tags in folksonomies

“…The proposed approach consists in, given an initial tag, deriving a set of semantically related tags and organizing the derived tags in a hierarchy according to mutual generalization and inclusion relationships. Cai et al [4] also adopt mutual overlapping measures to measure generalization and inclusion.…”

“…Cai et al [4] construct hierarchy of concepts based on the mutual-overlapping of tags. Similarly to what can be observed in Section 3.2.3 above, the mutual-overlapping also measures the fraction of resources tagged with two distinct tags.…”

“…Thus, one can assume that ti subsumes tj. Further assumptions about a and b values are detailed in [4].…”

“…To overcome this problem, many research works propose to construct hierarchies of tags automatically by means of heuristic algorithms [4,11,14]. In this paper, we propose a more principled approach where the subsumption relations are learned directly from folksonomy data.…”

Learning Structured Knowledge from Social Tagging Data: A Critical Review of Methods and Techniques

“…If two tags t i and t j co-occur, then the relationship "t i subsumes t j " can be defined if two conditions hold: first, t i labels more resources than t j ; second, the generalization degree of t i vs. t j is greater than a predefined threshold. More recently, a similar algorithm based on the resource set, mutual overlapping, is also proposed in [42]. It is demonstrated that mutual overlapping performs well on the dataset collected from the e-business website Taobao 10 .…”