SSTM 2015 Abstracts

Full Papers
Paper Nr: 1

Large Scale Web-Content Classification


Luca Deri, Maurizio Martinelli, Daniele Sartiano and Loredana Sideri

Abstract: Web classification is used in many security devices for preventing users to access selected web sites that are not allowed by the current security policy, as well for improving web search and for implementing contextual advertising. There are many commercial web classification services available on the market and a few publicly available web directory services. Unfortunately they mostly focus on English-speaking web sites, making them unsuitable for other languages in terms of classification reliability and coverage. This paper covers the design and implementation of a web-based classification tool for TLDs (Top Level Domain). Each domain is classified by analysing the main domain web site, and classifying it in categories according to its content. The tool has been successfully validated by classifying all the registered .it Internet domains, whose results are presented in this paper.

Paper Nr: 4

Non-negative Matrix Factorization for Binary Data


Jacob Søgaard Larsen and Line Katrine Harder Clemmensen

Abstract: We propose the Logistic Non-negative Matrix Factorization for decomposition of binary data. Binary data are frequently generated in e.g. text analysis, sensory data, market basket data etc. A common method for analysing non-negative data is the Non-negative Matrix Factorization, though this is in theory not appropriate for binary data, and thus we propose a novel Non-negative Matrix Factorization based on the logistic link function. Furthermore we generalize the method to handle missing data. The formulation of the method is compared to a previously proposed logistic matrix factorization without non-negativity constraint on the features. We compare the performance of the Logistic Non-negative Matrix Factorization to Least Squares Non-negative Matrix Factorization and Kullback-Leibler (KL) Non-negative Matrix Factorization on sets of binary data: a synthetic dataset, a set of student comments on their professors collected in a binary termdocument matrix and a sensory dataset. We find that choosing the number of components is an essential part in the modelling and interpretation, that is still unresolved.

Short Papers
Paper Nr: 3

Using the Cluster-based Tree Structure of k-Nearest Neighbor to Reduce the Effort Required to Classify Unlabeled Large Datasets


Elias Oliveira, Howard Roatti, Matheus de Araujo Nogueira, Henrique Gomes Basoni and Patrick Marques Ciarelli

Abstract: The usual practice in the classification problem is to create a set of labeled data for training and then use it to tune a classifier for predicting the classes of the remaining items in the dataset. However, labeled data demand great human effort, and classification by specialists is normally expensive and consumes a large amount of time. In this paper, we discuss how we can benefit from a cluster-based tree kNN structure to quickly build a training dataset from scratch. We evaluated the proposed method on some classification datasets, and the results are promising because we reduced the amount of labeling work by the specialists to 4% of the number of documents in the evaluated datasets. Furthermore, we achieved an average accuracy of 72.19% on tested datasets, versus 77.12% when using 90% of the dataset for training.

Paper Nr: 5

Performance Evaluation of Similarity Measures on Similar and Dissimilar Text Retrieval


Victor U. Thompson, Christo Panchev and Michael Michael Oakes

Abstract: Many Information Retrieval (IR) and Natural language processing (NLP) systems require textual similarity measurement in order to function, and do so with the help of similarity measures. Similarity measures function differently, some measures which work better on highly similar texts do not always do so well on highly dissimilar texts. In this paper, we evaluated the performances of eight popular similarity measures on four levels (degree) of textual similarity using a corpus of plagiarised texts. The evaluation was carried out in the context of candidate selection for plagiarism detection. Performance was measured in terms of recall, and the best performed similarity measure(s) for each degree of textual similarity was identified. Results from our Experiments show that the performances of most of the measures were equal on highly similar texts, with the exception of Euclidean distance and Jensen-Shannon divergence which had poorer performances. Cosine similarity and Bhattacharryan coefficient performed best on lightly reviewed text, and on heavily reviewed texts, Cosine similarity and Pearson Correlation performed best and next best respectively. Pearson Correlation had the best performance on highly dissimilar texts. The results also show term weighing methods and n-gram document representations that best optimises the performance of each of the similarity measures on a particular level of intertextual similarity.