Citation Context

...onducted extensive experiments in order to evaluate the performance of these naive approaches. As a second step, we have extended these strategies using estimators of mutual overlap among collections =-=[1]-=-. In Minerva, all queries are first processed by the query initiator itself on the locally available index. We expect that many queries will be answered this way without incurring any network costs. B...

Citation Context

...lgorithms using traditional single-term indexes in structured P2P networks do not scale to web size document collections. Similarly, even for more sophisticated schemes, such as term-to-peer indexing =-=[5, 4]-=- or hierarchical federated architectures [2, 9], there is little evidence on whether these approaches can scale to web sizes. The design of scalable models for full-text IR over P2P networks therefore...

Citation Context

...ng. The most important routing methods in the literature are CORI [13], simple forms of statistical language models [35, 30], the decision-theoretic framework (DTF) [21], and the overlap-aware method =-=[6, 33]-=-. All of the above methods organize the statistics about peers, which drive the query routing decisions, on a perterm basis and thus disregard term correlations. The only recent works that consider te...

Citation Context

...election problem involves selecting the best subset of sources for answering the query. Although source selection problem received some attention in the context of text and relational databases (c.f. =-=[26, 9, 13, 27, 21]-=-) existing approaches are focused on assessing the relevance of a source based on local measures of similarity between the query and the answers expected from the source. In the context of deep web, s...

Citation Context

...xtremely high overlap with regard to the own collection. The mutual overlap between peers has to be taken into account while selecting promising peers for a particular query. Overlap aware techniques =-=[5]-=- avoid retrieving a lot of redundant information so that a certain level of recall can be reached by querying fewer peers, compared to the non-overlap-aware approach. For instance, if the most promisi...

Citation Context

...he SPARQL endpoints’ data availability and run-time conditions. Overlap estimation among data sources have been used in a number of approaches in the area of distributed and P2P information retrieval =-=[4,10,15,18,23,24]-=-. COSCO [10] gathers statistics about coverage and overlap from past queries and uses them to determine in which order the overlapping collections should be accessed to retrieve the most new results i...

Citation Context

...neral background on P2P IR. Section 3 introduces the different types of synopses and presents our experimental comparison of the basic techniques. Section 4 introduces our P2P testbed, coined MINERVA =-=[5, 6]-=-. Section 5 develops the IQN routing method in detail. Section 6 discusses special techniques for handling multi-dimensional queries. Section 7 describes extensions to exploit histograms on score dist...

Citation Context

...the decision-theoretic framework [32], the GlOSS method [19], and methods based on statistical language models [37]. Recently, there has been research towards overlap aware resource selection methods =-=[20, 3, 30]-=- that consider the mutual overlap between peers during the selection process but do not consider global df estimation. Galanx [40] is a P2P search engine implemented using the Apache HTTP server and B...