Faceted product search powered by the Semantic Web Damir Vandic, Jan-Willem van Dam, Flavius Frasincar ⁎ Erasmus University Rotterdam, PO Box 1738, NL-3000 DR Rotterdam, The Netherlands abstract article info Article history: Received 17 June 2010 Received in revised form 17 October 2011 Accepted 15 February 2012 Available online 23 February 2012 Keywords: Semantic Web SPARQL and RDF Product identification Category mapping Product search This paper presents a platform for multifaceted product search using Semantic Web technology. Online shops can use a ping service to submit their RDFa annotated Web pages for processing. The platform is able to pro- cess these RDFa annotated (X)HTML pages and aggregate product information coming from different Web stores. We propose solutions for the identification of products and the mapping of the categories in this pro- cess. Furthermore, when a loose vocabulary such as the Google RDFa vocabulary is used, the platform deals with the issue of heterogeneous information (e.g., currencies, rating scales, etc.). © 2012 Elsevier B.V. All rights reserved. 1. Introduction Online product search, as a tool to help customers find their prod- ucts of interest, has become more important than ever as consumers nowadays purchase more often on the Web [1]. This is due to the fact that there is an increase in product specificity and consumer prefer- ence variation. The most important reason for this is technical ad- vancement, as this has led to a large increase of different product types. A second reason is that general wealth increase causes con- sumers to strengthen their preferences. The search space on the Web for products has also grown, which makes product search even more important. There are several problems with the current state of product search on the Web. First, the search engines cannot deal properly with synonyms and homonyms. Second, there is no good support for multiple languages, and more importantly, the aggregation of Web-wide information is seldom done. This is clear when we analyze the way we search for products on the Web. We keep switching back and forth from search results to find, for example, the cheapest price of a certain product. It would be useful if the product information is aggregated and shown to the user in one unified view. Third, there is no parametric Web-wide search available. Users cannot use queries like ‘all solar panels which give 12A output and cost less than $2000’. There are some localized, as opposed to Web-wide, product search Web sites where the user can perform this kind of parametric search. Usually these search engines only support basic product properties. Examples of these properties are the brand, the price, and the review rating of a product. Shopping.com, Google Products, and Shopzilla.com are three well-known parametric product search engines of such kind. A user can search, for example, for a washing machine with a max- imum price of £750 of the brand ‘Bosch’. Fig. 1 shows an example of this search. The user specifies the query constraints and the search engine queries the database, which contains all products, in order to display the washing machines that fulfill the requirements of the user. As a result of this, only stores that are indexed in the database of the search engine are shown. The databases of these kinds of search engines are updated through application programming interfaces (APIs) of Web shops that sell products. Of course, not every Web shop has an API and/or data feed possibilities. Furthermore, every search engine has its own standards which have to be obeyed by the Web shops. For instance, the API of Shopping.com is different than the API of Shopzilla. This means that not every Web shop will have their data prepared for both Shopping.com and Shopzilla. As it is costly to adjust data to a standard, it is not likely that a single search engine will receive data from all Web shops. By annotating Web pages with information on the Semantic Web, the APIs of nowadays can be made obsolete. The annotated information is also publicly available, which enables a search engine to gather product information directly from Web pages. There is one severe consequence of the current situation of prod- uct search. Because a user is not going to view all presented search re- sults, there is a chance that (s)he cannot precisely find a product that matches his or her criteria. What happens is that users more quickly start to focus on the price and give less weight to the product features. The result is that a fierce price competition arises. This can be consid- ered negative for both consumers and companies, as a user can prefer a product that meets all requirements but has a slightly higher price. Decision Support Systems 53 (2012) 425–437 ⁎ Corresponding author. E-mail addresses: vandic@ese.eur.nl (D. Vandic), info@qdentity.nl (J.W. van Dam), frasincar@ese.eur.nl (F. Frasincar). 0167-9236/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.dss.2012.02.010 Contents lists available at SciVerse ScienceDirect Decision Support Systems journal homepage: www.elsevier.com/locate/dss