Entity Search with NECESSITY

Loosely structured heterogeneous information spaces are typically created by merging data from a variety of different applications and information sources. A common problem these information spaces need to address is that various data describe the same real-word entities (e.g., people, conferences, organizations). In this demo, we introduce necessity, an efficient and scalable entity store. necessity is able to handle a large number of entities and at the same time provide an efficient and highly accurate entity search functionality for heterogeneous and partially structured queries that follow the vision of dataspaces. 1. MOTIVATION AND OUTLINE We are currently witnessing a rapid increase in the number of loosely structured heterogeneous information spaces collections of data coming from a variety of different applications and information sources. One common problem these information spaces face, is managing their entities (e.g., organizations, events), since there will be given various representations for the same real world entities. The necessity entity store is able to address this challenge. Our system can handle a large number of entities and at the same time provides an efficient and highly accurate entity search functionality. necessity stores entity profiles composed by a set of attributevalue pairs. It allows efficient entity search over these loosely structured heterogeneous information spaces, with queries being conditions on the entity’s attributes or values. Our approach contributes to the idea of realizing dataspaces as envisioned in [3]. This includes developing data models and designing search methods for a large collection of interrelated data, even if the entity queries are specific to our context. necessity was implemented and evaluated in the conThe name of the system was inspired by philosopher William of Ockham’s famous principle: “Entities must not be multiplied beyond necessity”. Copyright is held by the author/owner. Twelfth International Workshop on the Web and Databases (WebDB 2009), June 28, 2009, Providence, Rhode Island, USA. text of the Entity Name System (ENS), for the OKKAM project. The aim of ENS is to foster the global re-use of entity identifiers and to mediate between existing identifiers for individual entities (details available in [1]). ENS receives queries and checks whether the entity described in each query exists in OKKAM. If the entity exists OKKAM returns the corresponding identifier. The core benefit of ENS is to ease integration of external applications. For example, repositories from personal information management systems can now rely on the service provided by ENS for creating their URIs for entities. As such, the integrating challenge of knowing which representations in different repositories refer to the same entity, would be resolved by the use of shared IDs as issued by OKKAM. There are further application types that can profit from our proposed approach. One example is entity search in collaboratively authored information spaces, such as Wikipedia. Each Wikipedia entry is composed by various contributors who are not enforced to follow a specific format or schemata in a consistent way. Moreover, processing a human query over Wikipedia data could benefit from matching the query with the heterogeneous data of the entities. Another example of targeted applications for necessity is entity search engines. These applications are built upon information extracted from Web pages. Integrating this extracted data imposes a matching challenge of effectively identifying and merging the existing data that refer to the same real world entities. In addition, searching for a specific entity through this plethora of entities requires advanced matching functionalities. The rest of this paper is organized as follows. Section 2 presents the necessity entity store, with main focus on the entity search process. Section 3 describes the demo scenario, and Section 4 conclusions along with future work. 2. ENTITY SEARCH PROCESS Entities in necessity are modeled as a set of attributevalue pairs; a representation similar to dataspace proposal [3]. For example, a person entity will be represented by name, affiliation, email address, and whatever else is available. Entity search allows users or applications to retrieve the entities —ideally only one— already in necessity that best match an entity description provided as a query to the system. An entity query is a set of predicates, where each predicate is a keyword or an attribute value pair, e.g., Q1: name=“John Smith” EPFL, and Q2: name=Smith affiliation=EPFL. http://www.okkam.org http://www.wikipedia.org