Logo des Repositoriums

Auflistung nach Autor:in "Frommholz, Ingo"

1 - 4 von 4
  • Konferenzbeitrag
    Kontextbasiertes Retrieval unter Verwendung verknüpfter Annotationen
    (Informatik bewegt: Informatik 2002 - 32. Jahrestagung der Gesellschaft für Informatik e.v. (GI), 2002) Frommholz, Ingo; Brocks, Holger; Thiel, Ulrich; Stein, Adelheit
    Kollaborative Arbeitsumgebungen im Web können Mechanismen enthalten, mit denen neben dem Erstellen von zum Dokument gehörenden Metadaten auch ein wissenschaftlicher Diskurs über das eigentliche Dokument geführt werden kann (z.B. über freie Annotationen). Dieser Diskurs kann wertvolle Informationen über das Dokument enthalten, die aus den Metadaten nicht ersichtlich sind. Es wird gezeigt, wie sich ein solcher wissenschaftlicher Diskurs mittels Annotationen und Diskursstrukturrelationen modellieren läßt und wie man die daraus gewonnenen Informationen beim Retrieval ausnutzen kann.
  • Zeitschriftenartikel
    On Textual Analysis and Machine Learning for Cyberstalking Detection
    (Datenbank-Spektrum: Vol. 16, No. 2, 2016) Frommholz, Ingo; al-Khateeb, Haider M.; Potthast, Martin; Ghasem, Zinnar; Shukla, Mitul; Short, Emma
    Cyber security has become a major concern for users and businesses alike. Cyberstalking and harassment have been identified as a growing anti-social problem. Besides detecting cyberstalking and harassment, there is the need to gather digital evidence, often by the victim. To this end, we provide an overview of and discuss relevant technological means, in particular coming from text analytics as well as machine learning, that are capable to address the above challenges. We present a framework for the detection of text-based cyberstalking and the role and challenges of some core techniques such as author identification, text classification and personalisation. We then discuss PAN, a network and evaluation initiative that focusses on digital text forensics, in particular author identification.
  • Zeitschriftenartikel
    Scalable DB+IR Technology: Processing Probabilistic Datalog with HySpirit
    (Datenbank-Spektrum: Vol. 16, No. 1, 2016) Frommholz, Ingo; Roelleke, Thomas
    Probabilistic Datalog (PDatalog, proposed in 1995) is a probabilistic variant of Datalog and a nice conceptual idea to model Information Retrieval in a logical, rule-based programming paradigm. Making PDatalog work in real-world applications requires more than probabilistic facts and rules, and the semantics associated with the evaluation of the programs. We report in this paper some of the key features of the HySpirit system required to scale the execution of PDatalog programs.Firstly, there is the requirement to express probability estimation in PDatalog. Secondly, fuzzy-like predicates are required to model vague predicates (e.g. vague match of attributes such as age or price). Thirdly, to handle large data sets there are scalability issues to be addressed, and therefore, HySpirit provides probabilistic relational indexes and parallel and distributed processing. The main contribution of this paper is a consolidated view on the methods of the HySpirit system to make PDatalog applicable in real-scale applications that involve a wide range of requirements typical for data (information) management and analysis.
  • Zeitschriftenartikel
    The BCS Information Retrieval Specialist Group
    (Datenbank-Spektrum: Vol. 14, No. 1, 2014) Frommholz, Ingo; Kruschwitz, Udo; Tait, John