[Date Prev][Date Next][Date Index]

Talks on Recommender Systems and Data Extraction

Thursday, July 1st, Seminarroom 184/2

14:00 (s.t.) Cai Ziegler, Albert-Ludwigs University Freiburg:
"Crafting Decentralized Recommender Systems"

Endeavors investigating recommender systems have primarily addressed centralized scenarios and largely ignored open, decentralized systems, where remote information distribution and sparseness prevails. The absence of superordinate authorities having full access and control introduces serious issues necessitating novel approaches. Hence, our primary objective targets the design and conception of suchlike paradigms, relying upon two pivotal components, namely social network analysis and taxonomy-driven information filtering. Likewise applicable to other domains and scenarios, those two complementary pillars are seamlessly integrated into one coherent framework, eventually paving the way for decentralized recommenders.


15:00 Kai Simon, Albert-Ludwigs University Freiburg:
"Automatic Data Extraction (Crawling the hidden web)"

To address the problem of web data extraction, several techniques have been proposed in the literature. Using the taxonomy for characterizing web data extraction tools, the technique described in this talk falls into the "HTML-aware" category. Hereby the extraction rules are generated automatically without any learning or human intervention according to the DOM tree, and the tag token sequence, respectively. The main assumptions for this strategy follows the idea that the target pages were dynamically generated from data sources by one common schema and they contain data objects with a common HTML tag sequence. By analyzing several pages belonging to the same schema one tries to infer a regular expression which is able to extract the data of interest. In this talk we try to solve this problem with methods known from bioinformatics tailored for our purposes.