Recommending tags with a model of human categorization

Abstract

When interacting with social tagging systems, humans exercise complex processes of categorization that have been the topic of much research in cognitive science. In this paper we present a recommender approach for social tags derived from ALCOVE, a model of human category learning. The basic architecture is a simple three-layers connectionist model. The input layer encodes patterns of semantic features of a user-specific resource, such as latent topics elicited through Latent Dirichlet Allocation (LDA) or available external categories. The hidden layer categorizes the resource by matching the encoded pattern against already learned exemplar patterns. The latter are composed of unique feature patterns and associated tag distributions. Finally, the output layer samples tags from the associated tag distributions to verbalize the preceding categorization process. We have evaluated this approach on a real-world folksonomy gathered from Wikipedia bookmarks in Delicious. In the experiment our approach outperformed LDA, a well-established algorithm. We attribute this to the fact that our approach processes semantic information (either latent topics or external categories) across the three different layers. With this paper, we demonstrate that a theoretically guided design of algorithms not only holds potential for improving existing recommendation mechanisms, but it also allows us to derive more generalizable insights about how human information interaction on the Web is determined by both semantic and verbal processes.

Verbatim and Semantic Imitation in Indexing Resources on the Web: A Fuzzy‐trace Account of Social Tagging

“…For instance, in a recent work, Seitlinger, Kowald, Trattner, and Ley (2013) have emulated the gist-based production of tags as a simple connectionist model and demonstrated that a tag recommender based on this model provides more accurate recommendations than a well-established latent Dirichlet allocation approach.…”

Long time no see

“…For instance, in a recent study [28], we introduced a mechanism by which memory processes involved in tagging can be modeled on two levels of knowledge representation: on a semantic level (representing categories or LDA topics) and on a verbal level (representing tags). Next, we will aim at combining this integrative mechanism with the BLL equation to examine a potential interaction between the impact of recency (time-based forgetting) and the level of knowledge representation.…”

Attention Please! A Hybrid Resource Recommender Mimicking Attention-Interpretation Dynamics

“…We focused on social tagging systems for our study not only because their datasets are freely-available for scientific purposes but also because tagging data can be easily utilized to derive semantic topics for resources from it [17] by means of Latent Dirichlet Allocation (LDA) [25]. This initial step was necessary since our datasets did not contain categories or topics for the resources (e.g., like Wikipedia categories, see [35]) that are the basis for our approach. Thus, we chose LDA to simulate this external categorization.…”