STICS – Search and Analysis with Strings, Things, and Cats

Johannes Hoffart & Dragan Milchevski & Gerhard Weikum

STICS – Search and Analysis with Strings, Things, and Cats

„Things, not Strings” has been Google’s motto when introducing the Knowledge Graph and the entity-awareness of its search engine. When you type in the keyword “Klitschko” as a query, Google still returns Web and news pages, but also explicit entities like Wladimir Klitschko and his brother Vitali (including structured attributes like date of birth, profession, and relations to other entities, from the Knowledge Graph). Moreover, while typing, the query auto-completion method suggests the two brothers in entity form with the additional hints that one is an active boxer and the other a politician.

However, the Google approach still has limitations. First, recognizing entities in a keyword query and returning entity results seems to be limited to prominent entities. Unlike the Klitschko example, a query for the Ukrainian pop singer “Iryna Bilyk” does not show any entity suggestions (neither for auto-comple tion nor in the search results). Second, Google seems to understand only individual entities, but cannot handle sets of entities that are described by a type name or category phrase. For example, queries like “Ukrainian celebrities” or “East European politicians” return only the usual ten blue links: Web pages that match these phrases. The search engine does not understand the user’s intention to obtain lists of people in these categories.

STICS, short for “Searching with Strings, Things, and Cats”, is a novel search engine that extends entity awareness in Web and news searches by tapping into long-tail entities and understanding and expanding phrases that refer to semantic categories. STICS supports users in searching for strings, things, and cats (short for categories) in a seamless and convenient manner. For example, when posing the query “Merkel Ukrainian opposition”, the user is automatically guided, through auto- completion, to the entity ‘Angela_Merkel’ and the category ‘Ukrainian_politicians’, which is automatically expanded into ‘Vitali_Klitschko’, ‘Arseniy_Yatsenyuk’, etc. The search results include texts like “the German chancellor met the Ukrainian opposition leader and former heavy-weight champion”, even if these texts never mention the strings “Angela Merkel” and “Vitali Klitschko”. STICS achieves this by using the named entity recognition and disambiguation system AIDA, which links ambiguous words to entities in YAGO, which in turn contains the entities’ categories. The inner workings of AIDA are detailed in “AIDA – Resolving the Name Ambiguity”.