Citation Context

... there is poor support for advanced questions about factual knowledge. In this paper, we advocate the IQ paradigm for search against semantic knowledge bases or the linked-data cloud (linkeddata.org) =-=[3]-=- of structured data on the Web. We discuss the requirements for the exploration, filtering and aggregation steps, describe the development of effective tools in two case studies and identify challenge...

Citation Context

...ere is overlapping information in the knowledge-bases which can be leveraged) and subsequently, how to merge the (partial) results. Both tasks are non-trivial because on-the-fly entity disambiguation =-=[9]-=- may be required due to different vocabularies. With this overview of the IQ paradigm, we now illustrate two case studies of how a system built on these principles would work. In the first case study,...

Citation Context

...ema of an answer is the tree obtained by ignoring the names and contexts, that is, each node (and edge) has only a type. In this section, we consider the system demonstrated in [1], which is based on =-=[8]-=-. There are other systems for keywords search over data graph [11], but only [1] facilitates search in both dimensions, namely, answers and their schemas. Exploration. A major problem with typical key...

Citation Context

... bindings need to come from a single RDF triple. Fortunately, it is not too difficult to extend SPARQL, to support these cases while preserving the general flavor of SPARQL. Following the proposal by =-=[2]-=-, we could introduce variables that can be bound to entire paths in the RDF triples graph. Moreover, as we do not simply want connectivity but have semantic requirements, we would combine this with fi...

Citation Context

...nd make them queryable as if they were a fourth dimension added to the three SPO dimensions. Inspired by XQuery Full-Text, we have developed such a SPARQL extension with the following specific syntax =-=[6, 7]-=-: SELECT ?c, ?m WHERE { . . . ?c composed ?m{”classical music”} . } where “classical music” is a phrase to be matched in one or more of the witnesses for the triples that qualify for the triple patter...

Citation Context

...der to make user interactions as easy as possible, the system has to understand natural-language questions—a hard problem on which some progress has been made in the last few years (see, for example, =-=[10, 12]-=-). The questions need to be mapped to a query language in order to make the processing more precise and efficient, but there will be parts of the question which may not map to any structure. We believ...

Citation Context

...s and are moving towards more expressive entity-oriented search (e.g., by understanding product names or locations), the above questions cannot be answered easily. Recently emerging knowledge engines =-=[5, 4]-=- or knowledge-base search services such This article is published under a Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits distribution and reproductio...

Citation Context

...nd make them queryable as if they were a fourth dimension added to the three SPO dimensions. Inspired by XQuery Full-Text, we have developed such a SPARQL extension with the following specific syntax =-=[6, 7]-=-: SELECT ?c, ?m WHERE { . . . ?c composed ?m{”classical music”} . } where “classical music” is a phrase to be matched in one or more of the witnesses for the triples that qualify for the triple patter...

Citation Context

...for short. In fact, this is what searchengine users often end up doing, but there is not much support by the engine. Moreover, while IR researchers are advocating interactive retrieval for many years =-=[13]-=- (without compelling impact), in the structured-data world of knowledge bases and inference engines, the expectation by DB folks is that everything can be expressed in a single query of some super-pow...

Citation Context

...der to make user interactions as easy as possible, the system has to understand natural-language questions—a hard problem on which some progress has been made in the last few years (see, for example, =-=[10, 12]-=-). The questions need to be mapped to a query language in order to make the processing more precise and efficient, but there will be parts of the question which may not map to any structure. We believ...

Citation Context

...s and are moving towards more expressive entity-oriented search (e.g., by understanding product names or locations), the above questions cannot be answered easily. Recently emerging knowledge engines =-=[5, 4]-=- or knowledge-base search services such This article is published under a Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits distribution and reproductio...

Citation Context

...ber of answers. The schema of an answer is the tree obtained by ignoring the names and contexts, that is, each node (and edge) has only a type. In this section, we consider the system demonstrated in =-=[1]-=-, which is based on [8]. There are other systems for keywords search over data graph [11], but only [1] facilitates search in both dimensions, namely, answers and their schemas. Exploration. A major p...

Citation Context

...ontexts, that is, each node (and edge) has only a type. In this section, we consider the system demonstrated in [1], which is based on [8]. There are other systems for keywords search over data graph =-=[11]-=-, but only [1] facilitates search in both dimensions, namely, answers and their schemas. Exploration. A major problem with typical keyword search over data graphs is that the user may be inundated wit...