Logo des Repositoriums

Auflistung nach Autor:in "Zangerle, Eva"

1 - 3 von 3
  • Konferenzbeitrag
    Detection of Generated Text Reviews by Leveraging Methods from Authorship Attribution: Predictive Performance vs. Resourcefulness
    (BTW 2023, 2023) Moosleitner, Manfred; Specht, Günther; Zangerle, Eva
    Textual reviews are an integral part of online shopping and a source of information for potential customers. However, a prerequisite is that the reviews are authentic. To this end, pre-trained large language models have been shown to generate convincing text reviews at scale. Therefore, a critical task is the automatic detection of reviews not composed by a human, in a generated review classification task. State-of-the-art approaches to detect generated texts use pre-trained large language models, which exhibit hefty hardware requirements to run and fine-tune the model. Related work has shown that texts generated by language models often show differences in writing style and choice of words compared to texts written by humans. This two properties, which are unique per author, should be able to be utilized to identify if a text is generated by these algorithms. In this paper, we investigate the performance of features prominently used in authorship attribution tasks, using robust classifiers with substantially lower computational resources required. We show that features and methods from authorship attribution can be successfully applied for the task of detecting generated text reviews, leveraging the consistent writing style exhibited by large language models like GPT2. We argue that our approach achieves similar performance as state-of-the-art approaches while providing shorter training times and lower hardware requirements, necessary for, e.g, detection on the fly.
  • Konferenzbeitrag
    Pairwise Learning to Rank for Hit Song Prediction
    (BTW 2023, 2023) Mayerl, Maximilian; Vötter, Michael; Specht, Günther; Zangerle, Eva
    Predicting the popularity of songs in advance is of great interest to the music industry, with possible applications including assessing the potential of a new song, automated songwriting assistants, or song recommender systems. Traditional approaches for solving this use pointwise models focused on single songs, either using classification to categorize songs into classes like hit and non-hit, or regression to predict popularity metrics like play count. We propose to draw inspiration from research on learning to rank and instead use a pairwise model. Our model takes a pair of songs A and B and predicts whether song A is more popular than song B. Based on this problem formulation, we propose a neural network model that is trained in a pairwise fashion, as well as two data augmentation strategies for improving its performance. We also compare our model to one trained in a traditional pointwise way. Our results show that the pairwise model using our proposed augmentation strategies outperforms the pointwise model.
  • Zeitschriftenartikel
    SentiStorm: Echtzeit-Stimmungserkennung von Tweets
    (HMD Praxis der Wirtschaftsinformatik: Vol. 53, No. 4, 2016) Zangerle, Eva; Illecker, Martin; Specht, Günther
    Das automatisierte Erkennen der Stimmung von Texten hat in den letzten Jahren stark an Bedeutung gewonnen. Insbesondere durch die rapide Zunahme der Geschwindigkeit, mit der in sozialen Medien Informationen verbreitet werden, ist eine Echtzeit-Bestimmung der Stimmung von Texten ein herausforderndes Problem. Der Mikroblogging-Dienst Twitter verzeichnet im Durchschnitt über 8000 versendete Nachrichten pro Sekunde. In dieser Arbeit stellen wir mit dem SentiStorm-Ansatz einen Ansatz zur Stimmungserkennung von Tweets vor. Dabei erzeugen wir in einem ersten Schritt Merkmalsvektoren für die Tweets, die sowohl linguistische Informationen über den Tweet (Wichtigkeit der Wörter, Wortarten), wie auch über Sentiment-Lexika gewonnene Stimmungsinformationen beinhalten. In einem zweiten Schritt führen wir mittels der Merkmalsvektoren eine Stimmungsklassifikation durch, die eine Einteilung in positive, negative oder neutrale Tweets ermöglicht. Die durchgeführten Evaluationen zeigen, dass der präsentierte Ansatz bezüglich der Qualität der erkannten Stimmung sehr gute Erkennungsraten garantiert. Weiter zeigen wir, dass der Ansatz mittels der Apache Storm Plattform problemlos für die Echtzeit-Stimmungserkennung von Tweets skaliert werden kann.AbstractThe automatic detection of the sentiment of texts has become more and more important throughout the last years. Particularly, the rapid increase of the speed at which information is spread in social media makes real-time sentiment detection a challenging task. On the microblogging platform Twitter, more than 8,000 messages are sent every second. In this work, we present the SentiStorm approach, an approach for sentiment detection within tweets. We base the approach on feature vectors which contain linguistic information about the tweet content (weighting of words, word categories), as well as sentiment information which we gather based on sentiment lexica. Subsequently, we facilitate these feature vectors for a sentiment classification task which allows for distinguishing positive, negative and neutral tweets. Our conducted evaluations show that the proposed approach shows high classification accuracy. At the same time, we show that utilizing the Apache Storm platform we are able to easily scale the approach towards a real-time sentiment classification of tweets.