Logo des Repositoriums

Auflistung nach Schlagwort "Recommender Systems"

1 - 10 von 17
  • Zeitschriftenartikel
    ALEA – Adaptive eLEArning System – Lernende datenbankbasierte Lernsysteme in der Datenbanklehre
    (Datenbank-Spektrum: Vol. 21, No. 2, 2021) Schneider, Kerstin; Keller, Fabian; Habekost, Pia; Schmeil, Victor; Dampmann, Markus; Schorch, Florian
    Im Bereich Datenbanken werden selbstentwickelte E‑Learning-Systeme an vielen Hochschulen seit Jahren erfolgreich verwendet. An der Hochschule Harz werden E‑Learning-Systeme im Bereich Datenbanken im Rahmen der Lehre und für die Lehre entwickelt, weiterentwickelt und eingesetzt. Das Gesamtsystem, welches die zusammengehörenden Systeme umfasst, wird als ALEA bezeichnet. Es werden relevante Komponenten von ALEA erläutert, die im Rahmen der Datenbanklehre für die klassischen Teilgebiete SQL, ER-to-Relational-Mapping und Normalisierung genutzt werden.
  • Workshopbeitrag
    Audit, Don’t Explain – Recommendations Based on a Socio-Technical Understanding of ML-Based Systems
    (Mensch und Computer 2021 - Workshopband, 2021) Heuer, Hendrik
    In this position paper, I provide a socio-technical perspective on machine learning-based systems. I also explain why systematic audits may be preferable to explainable AI systems. I make concrete recommendations for how institutions governed by public law akin to the German TÜV and Stiftung Wartentest can ensure that ML systems operate in the interest of the public.
  • Textdokument
    Content-based Recommendations for Radio Stations with Deep Learned Audio Fingerprints
    (INFORMATIK 2020, 2021) Langer, Stefan; Obermeier, Liza; Ebert, André; Friedrich, Markus; Munisamy, Emma
    The world of linear radio broadcasting is characterized by a wide variety of stations and played content. That is why finding stations playing the preferred content is a tough task for a potential listener, especially due to the overwhelming number of offered choices. Here, recommender systems usually step in but existing content-based approaches rely on metadata and thus are constrained by the available data quality. Other approaches leverage user behavior data and thus do not exploit any domain-specific knowledge and are furthermore disadvantageous regarding privacy concerns. Therefore, we propose a new pipeline for the generation of audio-based radio station fingerprints relying on audio stream crawling and a Deep Autoencoder. We show that the proposed fingerprints are especially useful for characterizing radio stations by their audio content and thus are an excellent representation for meaningful and reliable radio station recommendations. Furthermore, the proposed modules are part of the HRADIO Communication Platform, which enables hybrid radio features to radio stations. It is released with a flexible open source license and enables especially small-and medium-sized businesses, to provide customized and high quality radio services to potential listeners.
  • Konferenzbeitrag
    Enhancing Explainability and Scrutability of Recommender Systems
    (BTW 2023, 2023) Ghazimatin, Azin
    Our increasing reliance on complex algorithms for recommendations calls for models and methods for explainable, scrutable, and trustworthy AI. While explainability is required for understanding the relationships between model inputs and outputs, a scrutable system allows us to modify its behavior as desired. These properties help bridge the gap between our expectations as end users and the algorithm's behavior and accordingly boost our trust in AI. Aiming to cope with information overload, recommender systems play a crucial role in filtering content (such as products, news, songs, and movies) and shaping a personalized experience for their users. Consequently, there has been a growing demand from the information consumers to receive proper explanations for their personalized recommendations. To this end, we put forward proposals for explaining recommendations to the end users. These explanations aim at helping users understand why certain items are recommended to them and how their previous inputs to the system relate to the generation of such recommendations. Such explanations usually contain valuable clues as to how a system perceives user preferences and more importantly how its behavior can be modified. Therefore, as a natural next step, we develop a framework for leveraging user feedback on explanations to improve their future recommendations. We evaluate all the proposed models and methods with real user studies and demonstrate their benefits at achieving explainability and scrutability in recommender systems.
  • Konferenzbeitrag
    Item Familiarity as a Possible Confounding Factor in User-Centric Recommender Systems Evaluation
    (i-com: Vol. 14, No. 1, 2015) Jannach, Dietmar; Lerche, Lukas; Jugovac, Michael
    User studies play an important role in academic research in the field of recommender systems as they allow us to assess quality factors other than the predictive accuracy of the underlying algorithms. User satisfaction is one such factor that is often evaluated in laboratory settings and in many experimental designs one task of the participants is to assess the suitability of the system-generated recommendations. The effort required by the user to make such an assessment can, however, depend on the user’s familiarity with the presented items and directly impact on the reported user satisfaction. In this paper, we report the results of a preliminary recommender systems user study using Mechanical Turk, which indicates that item familiarity is strongly correlated with overall satisfaction.
  • Zeitschriftenartikel
    Leveraging Arguments in User Reviews for Generating and Explaining Recommendations
    (Datenbank-Spektrum: Vol. 20, No. 2, 2020) Donkers, Tim; Ziegler, Jürgen
    Review texts constitute a valuable source for making system-generated recommendations both more accurate and more transparent. Reviews typically contain statements providing argumentative support for a given item rating that can be exploited to explain the recommended items in a personalized manner. We propose a novel method called Aspect-based Transparent Memories (ATM) to model user preferences with respect to relevant aspects and compare them to item properties to predict ratings, and, by the same mechanism, explain why an item is recommended. The ATM architecture consists of two neural memories that can be viewed as arrays of slots for storing information about users and items. The first memory component encodes representations of sentences composed by the target user while the second holds an equivalent representation for the target item based on statements of other users. An offline evaluation was performed with three datasets, showing advantages over two baselines, the well-established Matrix Factorization technique and a recent competitive representative of neural attentional recommender techniques.
  • Konferenzbeitrag
    Make Me Laugh: Recommending Humoristic Content on the WWW
    (Mensch und Computer 2015 – Proceedings, 2015) Buschek, Daniel; Just, Ingo; Fritzsche, Benjamin; Alt, Florian
    Humoristic content is an inherent part of the World Wide Web and increasingly consumed for micro-entertainment. However, humor is often highly individual and depends on background knowledge and context. This paper presents an approach to recommend humoristic content fitting each individual user's taste and interests. In a field study with 150 participants over four weeks, users rated content with a 0-10 scale on a humor website. Based on this data, we train and apply a Collaborative Filtering (CF) algorithm to assess individual humor and recommend fitting content. Our study shows that users rate recommended content 22.6% higher than randomly chosen content.
  • Konferenzbeitrag
    Merging Interactive Information Filtering and Recommender Algorithms – Model and Concept Demonstrator
    (i-com: Vol. 14, No. 1, 2015) Loepp, Benedikt; Herrmanny, Katja; Ziegler, Jürgen
    To increase controllability and transparency in recommender systems, recent research has been putting more focus on integrating interactive techniques with recommender algorithms. In this paper, we propose a model of interactive recommending that structures the different interactions users can have with recommender systems. Furthermore, as a novel approach to interactive recommending, we describe a technique that combines faceted information filtering with different algorithmic recommender techniques. We refer to this approach as blended recommending. We also present an interactive movie recommender based on this approach and report on its user-centered design process, in particular an evaluation study in which we compared our system with a standard faceted filtering system. The results indicate a higher level of perceived user control, more detailed preference settings, and better suitability when the search goal is vague.
  • Konferenzbeitrag
    MyMovieMixer: Ein hybrider Recommender mit visuellem Bedienkonzept
    (Mensch & Computer 2014 - Tagungsband, 2014) Herrmanny, Katja; Schering, Sandra; Berger, Ralf; Loepp, Benedikt; Günter, Timo; Hussein, Tim; Ziegler, Jürgen
    In diesem Beitrag stellen wir ein neuartiges, auf direkter Manipulation beruhendes Bedienkonzept für komplexe hybride Empfehlungssysteme anhand des von uns entwickelten Film-Recommenders MyMovieMixer vor. Der Ansatz ermöglicht es den Nutzern, ein hybrides Recommender-System mit einem komplexen Zusammenwirken verschiedener Filtermethoden durch interaktive und visuelle Methoden intuitiv zu steuern. Gleichzeitig wird die Transparenz der Empfehlungsgenerierung deutlich erhöht. Die Ergebnisse einer empirischen Evaluation des Systems zeigen, dass der Ansatz in Bezug auf Usability, User Experience, Intuitivität, Transparenz, wahrgenommene Empfehlungsqualität und somit letztlich im Hinblick auf die Nutzerzufriedenheit vielversprechend ist.
  • Konferenzbeitrag
    A Neural Natural Language Processing System for Educational Resource Knowledge Domain Classification
    (DELFI 2021, 2021) Schrumpf, Johannes; Weber, Felix; Thelen, Tobias
    In higher education, educational resources are the vessel with which information get transferred to the learner. Information on the content discussed in the scope of the educational resources, however, is implicit and must be inferred by the user by reading the resource title or through contextual information. In this paper we present a state-of-the-art neural natural language processing system, based on Google-BERT, that maps educational resource titles into one of 905 classes from the Dewey Decimal Classification (DDC) system. We present model architecture, training procedure dataset properties and our performance analysis methodology. We show that aside from classification performance, our model implicitly learns the class hierarchy inherent to the DDC.