Incorporating Multi-partite Networks and Expertise to Construct Related-Term Graphs

Abstract

Term suggestion techniques recommend query terms to a user based on his initial query. Providing adequate term suggestions is a challenging task. Most existing commercial search engines suggest search terms based on the frequency of prior used terms that match the first few letters typed by the user. We present a novel mechanism to construct semantic term-relation graphs to suggest semantically relevant search terms. We build term relation graphs based on multipartite networks of existing social media. These linkage networks are extracted from Wikipedia to eventually form term relation graphs. We propose incorporating contributorcategory networks to model the contributor expertise. This step has been shown to significantly enhance the accuracy of the inferred relatedness of the term-semantic graphs. Experiments showed the obvious advantage of our algorithms over existing approaches.

A Conceptualisation of Crowd Knowledge

“…In more tech-savvy articles, the CK system and its functions are described. Shieh et al (2010) regard social media as a breeding ground for CK and explicitly name Wikipedia. Furthermore, Erickson et al (2012) describe CK as an encapsulated artefact that can be integrated into products, while Shirazi et al (2010) use the current state of connected devices to form a situational CK snapshot that synchronises all devices and lets, for example, mobile phones become 'semi-intelligent' by automatically turning silent when entering a cinema.…”

“…Furthermore, it can be implemented in products, either static, for example as a source of training data to improve algorithms (Erickson et al, 2012), or dynamic, like the intelligent mobile phone (Shirazi et al, 2010). Discussions are automatically stored on the platform on which they operate (Schlagwein & Hu, 2016;Shieh et al, 2010). Centralised repositories are used, if CK is intentionally gathered, and often strongly linked to an environmental context (Erickson et al, 2012;Minguillón & Conesa, 2011).…”

“…Moreover, the type of storage influences how CK can be accessed. Repositories (Erickson et al, 2012;Minguillón & Conesa, 2011), social media (Schlagwein & Hu, 2016;Shieh et al, 2010) and Q&A platforms (de Souza, Campos, & Maia, 2014b) can be accessed by all or restricted to a selected group. Either way, the artefact must be deliberately retrieved and used.…”