Exploiting the wisdom of the crowds for characterizing and connecting heterogeneous resources

Abstract

Heterogeneous content is an inherent problem for crosssystem search, recommendation and personalization. In this paper we investigate differences in topic coverage and the impact of topicstopics in different kinds of Web services. We use entity extraction and categorization to create 'fingerprints' that allow for meaningful comparison. As a basis taxonomy, we use the 23 main categories of Wikipedia Category Graph, which has been assembled over the years by the wisdom of the crowds. Following a proof of concept of our approach, we analyze differences in topic coverage and topic impact. The results show many differences between Web services like Twitter, Flickr and Delicious, which reflect users' behavior and the usage of each system. The paper concludes with a user study that demonstrates the benefits of fingerprints over traditional textual methods for recommendations of heterogeneous resources.

Uma Abordagem para Identificação de Similaridade Entre Recursos Educacionais Utilizando Bases de Conhecimento Externas

“…Dados ligados podem ser usados para agregar valor ao conjunto de dados original (Araújo & De Souza, 2011), desenvolvendo-o com associação a outros conjuntos de dados conexos, além de ser possível utilizá-los de diferentes formas e para diferentes propósitos, como é o caso do Sindice (Oren et al, 2008). Além disso, essas bases externas de conhecimento tem sido utilizadas para apoiar tarefas, tais como anotação semântica (Mendes, Jakob, García-Silva, & Bizer, 2011), alinhamento de entidades (Jain, Hitzler, Sheth, Verma, & Yeh, 2010), identificação de contexto (Kawase, Siehndel, Pereira Nunes, Herder, & Nejdl, 2014), dentre outras. Neste trabalho fazemos uso do sistema de categorias da DBpedia para determinar a similaridade entre recursos educacionais anotados semanticamente.…”

The ARCOMEM Architecture for Social- and Semantic-Driven Web Archiving

“…We use the link and category information supplied by Wikipedia to define the topic and the expertise level inherent in certain terms. A more detailed description of the categorization process is given in [31]. For defining the expertise level of a given concept, we use the distance to the root within the Wikipedia category graph.…”

Clustering and Dataset Interlinking Recommendation in the Linked Open Data Cloud

“…A more generic approach to create profiles on the Web was presented by (KAWASE et al, 2014 In this thesis, we evaluated and adopted Kawase's technique, which demonstrated to be suitable to determine the topical domain of dataset communities. The drawback of Fetahu's approach in our scenario is the large number of categories assigned to a given dataset, which hinders the identification and selection of the most representative topics of a dataset and, consequently, of a community.…”

“…We will use in this paper the profiling technique described in (KAWASE et al, 2014), that generates profiles or fingerprints for textual resources. The method has five steps:…”

“…The profiling module processes the documents retrieved from a dataset and computes a description that characterizes the content stored in the dataset. DRX implements the technique described in (KAWASE et al, 2014), that generates dataset profiles or fingerprints from plain text.…”