Can the crowd be a source of information? Is it possible to receive useful hints from comments, blogs and product reviews? In the era of Web 2.0, people are allowed to give their opinion about everything such as movies, hotels, etc.. These reviews are social knowledge, that can be exploited to suggest possibly interesting items to other people. The goal of the Hints From the Crowd (HFC) project is to build a NoSQL database system for large collections of product reviews; the database is queried by expressing a natural language sentence; the result is a list of products ranked based on the relevance of reviews w.r.t. the natural language sentence. The best ranked products in the result list can be seen as the best hints for the user based on crowd opinions (the reviews). The HFC prototype has been developed to be independent of the particular application domain of the collected product reviews. Queries are performed by evaluating a text-based ranking metric for sets of reviews, specifically devised for this system; the metric evaluates the relevance of product reviews w.r.t. a natural language sentence (the query). We present the architecture of the system, the ranking metric and analyze execution times.
(2013). The Hints from the Crowd Project [conference presentation - intervento a convegno]. Retrieved from http://hdl.handle.net/10446/30266
The Hints from the Crowd Project
FOSCI, Paolo;PSAILA, Giuseppe;
2013-01-01
Abstract
Can the crowd be a source of information? Is it possible to receive useful hints from comments, blogs and product reviews? In the era of Web 2.0, people are allowed to give their opinion about everything such as movies, hotels, etc.. These reviews are social knowledge, that can be exploited to suggest possibly interesting items to other people. The goal of the Hints From the Crowd (HFC) project is to build a NoSQL database system for large collections of product reviews; the database is queried by expressing a natural language sentence; the result is a list of products ranked based on the relevance of reviews w.r.t. the natural language sentence. The best ranked products in the result list can be seen as the best hints for the user based on crowd opinions (the reviews). The HFC prototype has been developed to be independent of the particular application domain of the collected product reviews. Queries are performed by evaluating a text-based ranking metric for sets of reviews, specifically devised for this system; the metric evaluates the relevance of product reviews w.r.t. a natural language sentence (the query). We present the architecture of the system, the ranking metric and analyze execution times.File | Dimensione del file | Formato | |
---|---|---|---|
DEXA2013.pdf
Solo gestori di archivio
Descrizione: draft - bozza
Dimensione del file
334.48 kB
Formato
Adobe PDF
|
334.48 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo