Logo des Repositoriums

Auflistung nach Autor:in "Lofi, Christoph"

1 - 3 von 3
  • Zeitschriftenartikel
    Information Extraction Meets Crowdsourcing: A Promising Couple
    (Datenbank-Spektrum: Vol. 12, No. 2, 2012) Lofi, Christoph; Selke, Joachim; Balke, Wolf-Tilo
    Recent years brought tremendous advancements in the area of automated information extraction. But still, problem scenarios remain where even state-of-the-art algorithms do not provide a satisfying solution. In these cases, another aspiring recent trend can be exploited to achieve the required extraction quality: explicit crowdsourcing of human intelligence tasks. In this paper, we discuss the synergies between information extraction and crowdsourcing. In particular, we methodically identify and classify the challenges and fallacies that arise when combining both approaches. Furthermore, we argue that for harnessing the full potential of either approach, true hybrid techniques must be considered. To demonstrate this point, we showcase such a hybrid technique, which tightly interweaves information extraction with crowdsourcing and machine learning to vastly surpass the abilities of either technique.
  • Konferenzbeitrag
    Just ask a human? - controlling quality in relational similarity and analogy processing using the crowd
    (Datenbanksysteme für Business, Technologie und Web (BTW) 2013 - Workshopband, 2013) Lofi, Christoph
    Advancing semantically meaningful and human-centered interaction paradigms for large information systems is one of the central challenges of current information system research. Here, systems which can capture different notions of `similarity' between entities promise to be particularly interesting. While simple entity similarity has been addresses numerous times, relational similarity between entities and especially the closely related challenge of processing analogies remain hard to approach algorithmically due to the semantic ambiguity often involved in these tasks. In this paper, we will therefore employ human workers via crowd-sourcing to establish a performance baseline. Then, we further improve on this baseline by combining the feedback of multiple workers in a meaningful fashion. Due to the ambiguous nature of analogies and relational similarity, traditional crowd-sourcing quality control techniques are less effective and therefore we develop novel techniques paying respect to the intrinsic consensual nature of the task at hand. These works will further pave the way for building true hybrid systems with human workers and heuristic algorithms combining their individual strength.
  • Textdokument
    Perceptual Relational Attributes: Navigating and Discovering Shared Perspectives from User-Generated Reviews
    (BTW 2019, 2019) Lofi, Christoph; Valle Torre, Manuel; Ye, Mengmeng
    Effectively modelling and querying experience items like movies, books, or games in databases is challenging because these items are better described by their resulting user experience or perceived properties than by factual attributes. However, such information is often subjective, disputed, or unclear. Thus, social judgments like comments, reviews, discussions, or ratings have become a ubiquitous component of most Web applications dealing with such items, especially in the e-commerce domain. However, they usually do not play major role in the query process, and are typically just shown to the user. In this paper, we will discuss how to use unstructured user reviews to build a structured semantic representation of database items such that these perceptual attributes are (at least implicitly) represented and usable for navigational queries. Especially, we argue that a central challenge when extracting perceptual attributes from social judgments is respecting the subjectivity of expressed opinions. We claim that no representation consisting of only a single tuple will be sufficient. Instead, such systems should aim at discovering shared perspectives, representing dominant perceptions and opinions, and exploiting those perspectives for query processing.