Query rewriting has become a very prominent tool for efficiently implementing ontology-mediated querying in practice. The technique was originally introduced in the context of DL-Lite, but is now increasingly being used also for more expressive DLs. While rewritings are not guaranteed to exist beyond DL-Lite, the simple structure of ontologies that emerge from practical applications gives hope that non-existence of rewritings is a rare case.
The aim of the talk is to survey FO- and Datalog-rewriting of ontology-mediated queries in description logics beyond DL-Lite. It is structured into three parts. The first part is concerned with FO-rewritings in Horn-DLs such as EL, ELI, and Horn-SHI, the second part considers FO-rewritings in non-Horn-DLs such as ALC and ALCI, and the third part is about Datalog-rewritings in non-Horn DLs. In all three parts, I will try to emphasize useful characterizations of FO-rewritability, practically efficient algorithms for constructing rewritings, and relevant computational complexity results.
At last year’s DL workshop Alon Halevy told us about Web tables and how Google makes sense of tabular data on the Web together with Web knowledge graphs. Somewhat surprinsing, a still more unconquered area for Web data extraction seems to be the realm of Open Data: rather than extracting struc- tured data from the surface Web, another emerging source of data on the Web are lots of structured data sets being published openly on various Open Data Portals (e.g. http://www.publicdata.eu, http://data.gov.gr, http://data.gov.uk, http://www.data.gov, http://data.gv.at, http://open.wien.at/, just to name a few). However, despite already offering structured data, these Open Data por- tals often offer only limited search functionality, and intergrating and using Open Data from these portals involves various challenges, such as data quality problems, heterogeneity within metadata descriptions, dynamics, or lack of se- mantic descriptions of the data. Driven by a practical use case, the Open City Data pipeline project, in this talk we will report on experiences and obstacles for collecting and integrating Open Data across various data sets. We wil discuss how both methods from knowledge representation and reasoning as well as from statistics and data mining can be used to tackle some issues we encountered.
Modern search engine result pages often contain a mixture of results from structured and unstructured sources. Where such mixtures of structured and unstructured information are called for, the state-of-the-art is to organize complex search engine result pages around entities. Generating such a mixture of entity-oriented results in response to a traditional keyword query raises a number of interesting retrieval challenges. How do we link queries to entities? How do we identify different aspects of entities in cases where we are unsure about the user’s intent? How do we associate an entity with a topic that a user appears to be interested in? And how do we explain the relation between entities that are being presented as being similar or related?
In this area, a wide variety of complementary and competing proposed solu- tions exist. This talk provides a snapshot of current approaches to entity-focused search engine result pages, illustrates key developments using example, and outlines open questions and research opportunities.
The talk is based in part on joint work with Lars Buitinck, David Graus, Xinyi Li, Edgar Meij, Daan Odijk, Ridho Reinanda, Isaac Sijaranamual, Manos Tsagkias, Christophe Van Gysel, Nikos Voskarides, and Wouter Weerkamp.