ORCID Profile
0000-0002-5325-9724
Current Organisation
Federation University Australia
Does something not look right? The information on this page has been harvested from data sources that may not be up to date. We continue to work with information providers to improve coverage and quality. To report an issue, use the Feedback Form.
Publisher: Springer Science and Business Media LLC
Date: 07-02-2017
Publisher: Elsevier BV
Date: 05-2022
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Date: 07-2021
Publisher: Elsevier BV
Date: 02-2019
Publisher: Springer Singapore
Date: 2019
Publisher: Association for Computing Machinery (ACM)
Date: 05-05-2021
DOI: 10.1145/3431728
Abstract: Topic modeling is an unsupervised learning task that discovers the hidden topics in a collection of documents. In turn, the discovered topics can be used for summarizing, organizing, and understanding the documents in the collection. Most of the existing techniques for topic modeling are derivatives of the Latent Dirichlet Allocation which uses a bag-of-word assumption for the documents. However, bag-of-words models completely dismiss the relationships between the words. For this reason, this article presents a two-stage algorithm for topic modelling that leverages word embeddings and word co-occurrence. In the first stage, we determine the topic-word distributions by soft-clustering a random set of embedded n -grams from the documents. In the second stage, we determine the document-topic distributions by s ling the topics of each document from the topic-word distributions. This approach leverages the distributional properties of word embeddings instead of using the bag-of-words assumption. Experimental results on various data sets from an Australian compensation organization show the remarkable comparative effectiveness of the proposed algorithm in a task of document classification.
Publisher: Association for Computing Machinery (ACM)
Date: 31-03-2023
DOI: 10.1145/3510420
Abstract: This article proposes an improved Bayesian scheme by focusing on the region in which Bayesian may fail to correctly identify labels and improve classification performance by handling those errors. Bayesian method, as a probabilistic classifier, uses Bayes’ theorem to calculate the probability of an instance belonging to a class, where the class label with a maximum probability is assigned to the instance. In a spam detection problem, it can be considered that the prediction of the Bayesian classifier is weak when the probability obtained for classes spam and non-spam are close to each other. Therefore, we define a threshold to determine weak prediction against strong prediction. A hybrid strategy using a two-layer Bayesian approach is presented: basic Bayesian (BBayes) and corrected weak region Bayesian (CWRBayes), which are concerned with strong and weak predictions, respectively. Both techniques, BBayes and CWRBayes, have the same classification mechanism, but they use different feature selection mechanisms. The proposed methods are implemented and evaluated over two datasets of spam e-mails, and the results show that the proposed method has better performance than the baseline of the naïve Bayesian and some other Bayesian variants.
Publisher: Wiley
Date: 24-07-2019
DOI: 10.1111/COIN.12187
Location: Iran (Islamic Republic of)
No related grants have been discovered for Sattar Seifollahi.