Lorenzo Blanco

Probabilistic models to reconcile complex data from inaccurate data sources

Several techniques have been developed to extract and integrate data from web sources. However, web data are inherently imprecise and uncertain. This paper addresses the issue of characterizing the uncertainty of data extracted from a number of inaccurate sources. We develop a probabilistic model to compute a probability distribution for the extracted values, and the accuracy of the sources. Our model considers the presence of sources that copy their contents from other sources, and manages the misleading consensus produced by copiers. We extend the models previously proposed in the literature by working on several attributes at a time to better leverage all the available evidence. We also report the results of several experiments on both synthetic and real-life data to show the effectiveness of the proposed approach.

Redundancy-driven web data extraction and integration

A large number of web sites publish pages containing structured information about recognizable concepts, but these data are only partially used by current applications. Although such information is spread across a myriad of sources, the web scale implies a relevant redundancy. We present a domain independent system that exploits the redundancy of information to automatically extract and integrate data from the Web. Our solution concentrates on sources that provide structured data about multiple instances from the same conceptual domain, e.g., financial data, product information. Our proposal is based on an original approach that exploits the mutual dependency between the data extraction and the data integration tasks. Experiments on a sample of 175,000 pages confirm the feasibility and quality of the approach.

Future Locations Prediction with Uncertain Data

The ability to predict future movements for moving objects enables better decisions in terms of time, cost, and impact on the environment. Unfortunately, future location prediction is a challenging task. Existing works exploit techniques to predict a trip destination, but they are effective only when location data are precise (e.g., GPS data) and movements are observed over long periods of time (e.g., weeks). We introduce a data mining approach based on a Hidden Markov Model (HMM) that overcomes these limits and improves existing results in terms of precision of the prediction, for both the route (i.e., trajectory) and the final destination. The model is resistant to uncertain location data, as it works with data collected by using cell-towers to localize the users instead of GPS devices, and reaches good prediction results in shorter times (days instead of weeks in a representative real-world application). Finally, we introduce an enhanced version of the model that is orders of magnitude faster than the standard HMM implementation.

Supporting the automatic construction of entity aware search engines

Several web sites deliver a large number of pages, each publishing data about one instance of some real world entity, such as an athlete, a stock quote, a book. Although it is easy for a human reader to recognize these instances, current search engines are unaware of them. Technologies for the Semantic Web aim at achieving this goal; however, so far they have been of little help in this respect, as semantic publishing is very limited. We have developed a method to automatically search on the web for pages that publish data representing an instance of a certain conceptual entity. Our method takes as input a small set of sample pages: it automatically infers a description of the underlying conceptual entity and then searches the web for other pages containing data representing the same entity. We have implemented our method in a system prototype, which has been used to conduct several experiments that have produced interesting results.

Flint: Google-basing the Web

Several Web sites deliver a large number of pages, each publishing data about one instance of some real world entity, such as an athlete, a stock quote, a book. Even though it is easy for a human reader to recognize these instances, current search engines are unaware of them. Technologies for the Semantic Web aim at achieving this goal; however, so far they have been of little help in this respect, as semantic publishing is very limited. We have developed a system, called Flint, for automatically searching, collecting and indexing Web pages that publish data representing an instance of a certain conceptual entity. Flint takes as input a small set of labeled sample pages: it automatically infers a description of the underlying conceptual entity and then searches the Web for other pages containing data representing the same entity. Flint automatically extracts data from the collected pages and stores them into a semi-structured self-describing database, such as Google Base. Also, the collected pages can be used to populate a custom search engine; to this end we rely on the facilities provided by Google Co-op.

Structure and Semantics of Data-IntensiveWeb Pages: An Experimental Study on their Relationships

In data-intensive web sitespages aregenerated by scriptsthat embed data from aback- end database into HTML templates. There is usually a relationship between the semantics of the data in a page and its corresponding template. For example, in a web site about sports events, it is likely that pages with data about athletes are associated with a template that differs from the template used to generate pages about coaches or referees. This article presents a method to classify web pages according to the associated template. Given a web page, the goal of our method is to accurately find the pages that are about the same topic. Our method leverages on a simple, yet effective model to abstract some structural features of a web page. We present the results of an extensive experimental analysis that show the performance of our methods in terms of both recall and precision regarding a large number of real-world web pages. same web site, but they provideinformationabout a football match, a football team, and some news. Although these pages have some common parts, e.g., headers and footers, the differencein their contents lead to substantial differencesin the HTMLpresentation. It is reasonable to assume that player pages are generated by a template that is different from the template used to generate match or team pages. In general, since each HTML template is tailored to organise the results of a specific query, it is reasonable to expect that there exists a correlation between the structure of a page and its semantics. This article investigates the effectiveness of template-based techniques for classify- ing web pages. We present a framework to model and compare the HTML structure of web pages, and provethat our techniques are effective enough by means of an extensive experimental study. Our proposal has many applications, namely: given a large wrapper library to extract and, possibly, integrate data from a large number of web sites, our techniques can be used to select the right wrapper in the library; furthermore, it can

Exploiting information redundancy to wring out structured data from the web

A large number of web sites publish pages containing structured information about recognizable concepts, but these data are only partially used by current applications. Although such information is spread across a myriad of sources, the web scale implies a relevant redundancy. We present a domain independent system that exploits the redundancy of information to automatically extract and integrate data from the Web. Our solution concentrates on sources that provide structured data about multiple instances from the same conceptual domain, e.g. financial data, product information. Our proposal is based on an original approach that exploits the mutual dependency between the data extraction and the data integration tasks. Experiments confirmed the quality and the feasibility of the approach.

Web data reconciliation: models and experiences

An increasing number of web sites offer structured information about recognizable concepts, relevant to many application domains, such as finance, sport, commercial products. However, web data is inherently imprecise and uncertain, and conflicting values can be provided by different web sources. Characterizing the uncertainty of web data represents an important issue and several models have been recently proposed in the literature. This chapter illustrates state-of-the-art Bayesan models to evaluate the quality of data extracted from the Web and reports the results of an extensive application of the models on real life web data. Experimental results show that for some applications even simple approaches can provide effective results, while sophisticated solutions are needed to obtain a more precise characterization of the uncertainty.

Characterizing the uncertainty of web data: models and experiences

An increasing number of web sites offer structured information about recognizable concepts, relevant to many application domains, such as finance, sport, commercial products. However, web data is inherently imprecise and uncertain, and conflicting values can be provided by different web sources. Characterizing the uncertainty of web data represents an important issue and several models have been recently proposed in the literature. The paper illustrates state-of-the-art Bayesan models to evaluate the quality of data extracted from the Web and reports the results of an extensive application of the models on real life web data. Our experimental results show that for some applications even simple approaches can provide effective results, while sophisticated solutions are needed to obtain a more precise characterization of the uncertainty.

Automatically building probabilistic databases from the web

A relevant number of web sites publish structured data about recognizable concepts (such as stock quotes, movies, restau- rants, etc.). There is a great chance to create applications that rely on a huge amount of data taken from the Web. We present an automatic and domain independent system that performs all the steps required to benefit from these data: it discovers data intensive web sites containing information about an entity of interest, extracts and integrate the published data, and finally performs a probabilistic analysis to characterize the impreciseness of the data and the accuracy of the sources. The results of the processing can be used to populate a probabilistic database.