Recommender Systems and Exploratory Search

Users tend to be overwhelmed by the massive amount of information available. Recommendation in online web services have gained momentum over the past years as a key factor to deliver personalized content and enhance the user experience. Reducing the information overload and assisting customers to make decision become part of primary concerns in the e-service area. In this sense, recommender systems attempt to provide efficient filters that decode the users' interests, and optimize accordingly the information perceived. To help these systems predict items of interest, various clues are available, ranging from a user profile, explicit ratings, to past activities and social interactions.

We research novel hybrid approaches that leverage Semantic Web technologies. On the one hand, we develop content-based recommendation systems enriched with Linked Data to overcome the data sparsity. On the other hand, we incorporate collaborative filtering to involve the social aspect, an influential feature in decision making. We propose to integrate a user diversity model designed to detect user propensity towards specific topics.

Exploring and Recommending Music

Music is everywhere − played, recorded, broadcasted and listened to. Files of recorded music are all over the web − stored, streamed, shared or sold. The knowledge about musical works is described in detail in the information systems of several cultural and media organizations around the world (e.g. the Bibliothèque Nationale de France, the Philharmonie de Paris, Radio France). We are working with those institutions to make this knowledge available and re-usable on the web of data.

Musical works are complex objects. Expressing them comprehensively requires the description of their physical manifestations (recordings, scores) and all the events that define them (creation, publication, performance). The first aspect is relatively well-mastered, in library catalogues as well as in the music industry. Several models can be used to describe a musical work, some being specific to the music domain (MusicOntology), while others being more broadly designed for libraries in general (FRBR, UNIMARC). The second aspect is rather new, although there is a growing need and interest in using even-based models. Several ontologies aim specifically to define events (Event, LODE), but there are fewer examples of using them to describe the creation and publication processes.

One of the difficulties with musical works is that although their expressions may differ significantly, they are still regarded as a single work. Modeling it requires to express the singleness of the work as well as the specificities of its expressions, and to show how events are connected to these expressions. Another issue is that an arrangement may be considered as an expression or as a new work, depending on the data producer. Therefore, we aim to propose an expressive ontological model that enables to represent music metadata in all its complexity, for example, starting from the CRM and FRBRoo ontologies. Furthermore, we are researching how to map complex models to a simpler one useful for web publication such as Schema.org.

Modeling a classical musical work as well as its expressions

How to visualize and interact with enriched music catalogs? How to design and implement an effective exploratory search engine that uses the rich semantic model for describing music? How to design a tool that can support the selection of music work, for example for recommending a program for a specialized radio channel? How to select music works for illustrating an historical period? or a movement or school? Music recommendations have been deeply studied by content holders such as Spotify, Deezer, Last.fm and Youtube among others. In this work, we study various recommendation strategies, that fully uses expert knowledge, words-of-mouth (collaborative filtering) and the rich metadata coming from the cultural institutions for different type of audience: experts in music or amateurs that incrementally build his music culture and identity.

As first results, we have developed Overture, a web application enabling to explore interlinked catalogs. Overture is an exploratory search engine prototype that enables to browse through the reconciled collection of bibliographical records of classical music and to highlight the various interpretations of a work, its derivative, its performance casting as well as other rich metadata.

Beyond the exploration of the catalogs, we also aim to develop novel recommendation algorithms that fully take into the richness of the semantic graph describing musical works. We have developed AR&TREx (ARtists and TRacks Explorer), a tool that enables the user to ask for recommendations based on a seed (song) to both the Spotify and the Last.fm recommendation APIs.

Exploratory Search

Contrary to lookup search engines that help users to retrieve specific items (e.g., names, numbers, short statements, or specific documents), Exploratory Search Systems (ESSs) are search engines that help users to explore a topic of interest. Exploratory search (ES) tasks are open-ended, multi-faceted, and iterative like learning or topic investigation. Currently, the evaluation methods of ESSs are not entirely adapted to the special features of ES tasks, and do not effectively assess that ESSs support users in performing those tasks. Our research goal is to elaborate methods that effectively lead to this assessment. In particular, we are designing a user-centered evaluation method of exploratory search systems based on a model of the exploratory search process. This research is conducted in collaboration with the INRIA Wimmics team.

Syndicate

Syndicate content

Data Science