Learning to Rank

New general purpose ranking functions are discovered using genetic programming. The TREC WSJ collection was chosen as a training set. A baseline comparison function was chosen as the best of inner product, probability, cosine, and Okapi BM25. An elitist genetic algorithm with a population size 100 was run 13 times for 100 generations and the best performing algorithms chosen from these. The best learned functions, when evaluated against the best baseline function (BM25), demonstrate some significant performance differences, with improvements in mean average precision as high as 32% observed on one TREC collection not used in training. In no test is BM25 shown to significantly outperform the best learned function.     

Semantic Similarity Search on Semistructured Data with the XXL Search Engine

Query languages for XML such as XPath or XQuery support Boolean retrieval: a query result is a (possibly restructured) subset of XML elements or entire documents that satisfy the search conditions of the query. This search paradigm works for highly schematic XML data collections such as electronic catalogs. However, for searching information in open environments such as the Web or intranets of large corporations, ranked retrieval is more appropriate: a query result is a ranked list of XML elements in descending order of (estimated) relevance. Web search engines, which are based on the ranked retrieval paradigm, do, however, not consider the additional information and rich annotations provided by the structure of XML documents and their element names.This article presents the XXL search engine that supports relevance ranking on XML data. XXL is particularly geared for path queries with wildcards that can span multiple XML collections and contain both exact-match as well as semantic-similarity search conditions. In addition, ontological information and suitable index structures are used to improve the search efficiency and effectiveness. XXL is fully implemented as a suite of Java classes and servlets. Experiments in the context of the INEX benchmark demonstrate the efficiency of the XXL search engine and underline its effectiveness for ranked retrieval.     

The Importance of Length Normalization for XML Retrieval

XML retrieval is a departure from standard document retrieval in which each individual XML element, ranging from italicized words or phrases to full blown articles, is a retrievable unit. The distribution of XML element lengths is unlike what we usually observe in standard document collections, prompting us to revisit the issue of document length normalization. We perform a comparative analysis of arbitrary elements versus relevant elements, and show the importance of element length as a parameter for XML retrieval. Within the language modeling framework, we investigate a range of techniques that deal with length either directly or indirectly. We observe a length-bias introduced by the amount of smoothing, and show the importance of extreme length bias for XML retrieval. We also show that simply removing shorter elements from the index (by introducing a cut-off value) does not create an appropriate element length normalization. Even after restricting the minimal size of XML elements occurring in the index, the importance of an extreme explicit length bias remains.     

Accessibility requirements take on new significance for publishers

Conclusion Publishers may wish to ignore disabilities legislation since it doesn’t apply directly to them, but their customers will not be able ignore the new laws when making purchasing decisions. Thus, as a practical matter each publisher must develop a game plan for how it will position itself to remain competitive in a marketplace where the timely availability of their print works in digital formats is becoming a Sine qua non on purchasing checklists. How individual publishers choose to tackle the new challenges will be determined by a number of variables including the type of content they publish (humanities, math, science, technical), the production systems they use, the length of their press runs, the shelf-life of their products, the level of service they are prepared to provide, the risks they are willing to take, and the dynamics of the markets they serve. A publisher’s strategy will need frequent review and modification because the variables involved in the accessibility environment are changing rapidly. New developments are occurring all the time in the legal environment, in production technologies, in assistive technologies, in distribution technologies, in digital rights management technologies, as well as in file standards. Solutions will need frequent tuning. For publishers, accessibility challenges do not stop with providing alternate formats for printed works. Those delivering their content over private networks via course management systems and over the Internet find that there are other accessibility requirements and standards to be met. It will behoove publishers to develop and implement informed operating policies and protocols that assure that onan ongoing basis its products and services meet applicable accessibility requirements and thus can fully compete in an increasingly demanding marketplace.