Journal of Intelligent & Fuzzy Systems - Volume 39, issue 2

Authors: González, José Ángel | Segarra, Encarna | García-Granada, Fernando | Sanchis, Emilio | Hurtado, Lluís-F.

Article Type: Research Article

Abstract: In this paper, we present an extractive approach to document summarization, the Siamese Hierarchical Transformer Encoders system, that is based on the use of siamese neural networks and the transformer encoders which are extended in a hierarchical way. The system, trained for binary classification, is able to assign attention scores to each sentence in the document. These scores are used to select the most relevant sentences to build the summary. The main novelty of our proposal is the use of self-attention mechanisms at sentence level for document summarization, instead of using only attentions at word level. The experimentation carried out …using the CNN/DailyMail summarization corpus shows promising results in-line with the state-of-the-art. Show more

Keywords: Siamese neural networks, self attention, extractive summarization

DOI: 10.3233/JIFS-179901

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2409-2419, 2020

Authors: Mendoza, Griselda Areli Matias | Ledeneva, Yulia | García-Hernández, Rene Arnulfo

Article Type: Research Article

Abstract: The methods of Automatic Extractive Summarization (AES) uses the features of the sentences of the original text to extract the most important information that will be considered in summary. It is known that the first sentences of the text are more relevant than the rest of the text (this heuristic is called baseline), so the position of the sentence (in reverse order) is used to determine its relevance, which means that the last sentences have practically no possibility of being selected. In this paper, we present a way to soften the importance of sentences according to the position. The comprehensive …tests were done on one of the best AES methods using the bag of words and n-grams models with the with DUC02 and DUC01 data sets to determine the importance of sentences. Show more

Keywords: Automatic Text Summarization, n-gram Model, bag of words model, slope calculation, genetic algorithm

DOI: 10.3233/JIFS-179902

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2421-2431, 2020

Authors: Céspedes-Hernández, David | González-Calleros, Juan Manuel | Guerrero-García, Josefina | Vanderdonckt, Jean

Article Type: Research Article

Abstract: A gesture elicitation study consists of a popular method for eliciting a sample of end end users to propose gestures for executing functions in a certain context of use, specified by its users and their functions, the device or the platform used, and the physical environment in which they are working. Gestures proposed in such a study needs to be classified and, perhaps, extended in order to feed a gesture recognizer. To support this process, we conducted a full-body gesture elicitation study for executing functions in a smart home environment by domestic end users in front of a camera. Instead …of defining functions opportunistically, we define them based on a taxonomy of abstract tasks. From these elicited gestures, a XML-compliant grammar for specifying resulting gestures is defined, created, and implemented to graphically represent, label, characterize, and formally present such full-body gestures. The formal notation for specifying such gestures is also useful to generate variations of elicited gestures to be applied on-the-fly on gestures in order to allow one-shot learning. Show more

Keywords: Gesture elicitation study, gesture grammar, gesture recognition, gesture user interfaces, engineering interactive computing systems, one-shot learning

DOI: 10.3233/JIFS-179903

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2433-2444, 2020

Authors: Shafiq, Hafiz Muhammad | Tahir, Bilal | Mehmood, Muhammad Amir

Article Type: Research Article

Abstract: Urdu is the most popular language in Pakistan which is spoken by millions of people across the globe. While English is considered the dominant web content language, characteristics of Urdu language web content are still unknown. In this paper, we study the World-Wide-Web (WWW) by focusing on the content present in the Perso-Arabic script. Leveraging from the Common Crawl Corpus, which is the largest publicly available web content of 2.87 billion documents for the period of December 2016, we examine different aspects of Urdu web content. We use the Compact Language Detector (CLD2) for language detection. We find that the …global WWW population has a share of 0.04% for Urdu web content with respect to document frequency. 70.9% of the top-level Urdu domains consist of . com , . org , and . info . Besides, urdulughat is the most dominating second-level domain. 40% of the domains are hosted in the United States while only 0.33% are hosted within Pakistan. Moreover, 25.68% web-pages have Urdu as primary language and only 11.78% of web-pages are exclusively in Urdu. Our Urdu corpus consists of 1.25 billion total and 18.14 million unique tokens. Furthermore, the corpus follows the Zipf’s law distribution. This Urdu Corpus can be used for text summarization, text classification, and cross-lingual information retrieval. Show more

Keywords: Urdu web corpus, Perso-Arabic script, web content analysis, common crawl corpus

DOI: 10.3233/JIFS-179904

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2445-2455, 2020

Authors: Amjad, Maaz | Sidorov, Grigori | Zhila, Alisa | Gómez-Adorno, Helena | Voronkov, Ilia | Gelbukh, Alexander

Article Type: Research Article

Abstract: The paper presents a new corpus for fake news detection in the Urdu language along with the baseline classification and its evaluation. With the escalating use of the Internet worldwide and substantially increasing impact produced by the availability of ambiguous information, the challenge to quickly identify fake news in digital media in various languages becomes more acute. We provide a manually assembled and verified dataset containing 900 news articles, 500 annotated as real and 400, as fake, allowing the investigation of automated fake news detection approaches in Urdu. The news articles in the truthful subset come from legitimate news sources, …and their validity has been manually verified. In the fake subset, the known difficulty of finding fake news was solved by hiring professional journalists native in Urdu who were instructed to intentionally write deceptive news articles. The dataset contains 5 different topics: (i) Business, (ii) Health, (iii) Showbiz, (iv) Sports, and (v) Technology. To establish our Urdu dataset as a benchmark, we performed baseline classification. We crafted a variety of text representation feature sets including word n -grams, character n -grams, functional word n -grams, and their combinations. After applying a variety of feature weighting schemes, we ran a series of classifiers on the train-test split. The results show sizable performance gains by AdaBoost classifier with 0.87 F1Fake and 0.90 F1Real . We provide the results evaluated against different metrics for a convenient comparison of future research. The dataset is publicly available for research purposes. Show more

Keywords: Fake news detection, urdu corpus, language resources, benchmark dataset, classification, machine learning

DOI: 10.3233/JIFS-179905

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2457-2469, 2020

Authors: Singh, Prashasti | Piryani, Rajesh | Singh, Vivek Kumar | Pinto, David

Article Type: Research Article

Abstract: Classification of research articles into different subject areas is an extremely important task in bibliometric analysis and information retrieval. There are primarily two kinds of subject classification approaches used in different academic databases: journal-based (aka source-level) and article-based (aka publication-level). The two popular academic databases- Web of Science and Scopus- use journal-based subject classification scheme for articles, which assigns articles into a subject based on the subject category assigned to the journal in which they are published. On the other hand, the recently introduced Dimensions database is the first large academic database that uses article-based subject classification scheme that assigns …the article to a subject category based on its contents. Though the subject classification schemes of Web of Science have been compared in several studies, no research studies have been done on comparison of the article-based and journal-based subject classification systems in different academic databases. This paper aims to compare the accuracy of subject classification system of the three popular academic databases: Web of Science, Scopus and Dimensions through a large-scale user-based study. Results show that the commonly held belief of superiority of article-based subject classification over the journal-based subject classification scheme does not hold at least at the moment, as Web of Science appears to have the most accurate subject classification. Show more

Keywords: Academic databases, research category, subject classification

DOI: 10.3233/JIFS-179906

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2471-2476, 2020

Authors: Karmakar, Mousumi | Singh, Vivek Kumar | Pinto, David

Article Type: Research Article

Abstract: With evolution of knowledge disciplines and cross fertilization of ideas, research outputs reported as scientific papers are now becoming more and more interdisciplinary. An interdisciplinary research work usually involves ideas and approaches from multiple disciplines of knowledge applied to solve a specific problem. In many cases the interdisciplinary areas eventually emerge as full-fledged disciplines. In the last two decades, several approaches have been proposed to measure the Interdisciplinarity of a scientific article, such as propositions based on authorship, references, set of keywords etc. Among all these approaches, reference-set based approach is most widely used. The diversity of knowledge in the …reference set has been measured with three parameters, namely variety , balance , and disparity . Different studies tried to combine these measures in one way or other to propose an aggregate measure of interdisciplinarity, called integrated diversity . However, there is a lack of understanding on inter-relations between these parameters. This paper tries to look into inter-relatedness between the three parameters by analytical study on an important interdisciplinary research area, Internet of Things (IoT). Research articles in IoT, as obtained from Web of Science for the year 2018 have been analyzed to compute the three measures and understand their inter-relatedness. Results obtained show that variety and balance are negatively correlated, variety and disparity do not show a stable relatedness and balance and disparity are negatively correlated. Further, the integrated diversity measure is negatively correlated with variety and weakly positively correlated with balance and disparity . The results imply that the composite integrated diversity measure may not be a suitably constructed composite measure of interdisciplinarity. Show more

Keywords: Diversity, interdisciplinarity, interdisciplinary research, multidisciplinary research

DOI: 10.3233/JIFS-179907

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2477-2485, 2020

Authors: Alekseev, Anton | Tutubalina, Elena | Malykh, Valentin | Nikolenko, Sergey

Article Type: Research Article

Abstract:  Deep learning architectures based on self-attention have recently achieved and surpassed state of the art results in the task of unsupervised aspect extraction and topic modeling. While models such as neural attention-based aspect extraction (ABAE) have been successfully applied to user-generated texts, they are less coherent when applied to traditional data sources such as news articles and newsgroup documents. In this work, we introduce a simple approach based on sentence filtering in order to improve topical aspects learned from newsgroups-based content without modifying the basic mechanism of ABAE. We train a probabilistic classifier to distinguish between out-of-domain texts (outer dataset) …and in-domain texts (target dataset). Then, during data preparation we filter out sentences that have a low probability of being in-domain and train the neural model on the remaining sentences. The positive effect of sentence filtering on topic coherence is demonstrated in comparison to aspect extraction models trained on unfiltered texts. Show more

Keywords: Aspect extraction, out-of-domain classification, deep learning, topic models, topic coherence

DOI: 10.3233/JIFS-179908

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2487-2496, 2020

Authors: Duchanoy, Carlos A. | Moreno-Armendáriz, Marco A. | Calvo, Hiram | Hernández-Ramos, Víctor E.

Article Type: Research Article

Abstract: LinkedIn is a social medium oriented to professional career handling and networking. In it, users write a textual profile on their experience, and add skill labels in a free format. Users are able to apply for different jobs, but specific feedback on the appropriateness of their application according to their skills is not provided to them. In this work we particularly focus on applicants of the project management branch from information technologies—although the presented methodology could be extended to any area following the same mechanism. Using the information users provide in their profile, it is possible to establish the corresponding …level in a predefined Project Manager career path (PM level). 1500+ experiences and skills from 300 profiles were manually tagged to train and test a model to automatically estimate the PM level. In this proposal we were able to perform such prediction with a precision of 98%. Additionally, the proposed model is able to provide feedback to users by offering a guideline of necessary skills to be learned to fulfill the current PM level, or those needed in order to upgrade to the following PM level. This is achieved through the clustering of skill qualification labels. Results of experiments with several clustering algorithms are provided as part of this work. Show more

Keywords: Project manager career path level, profile classification, skill qualification estimation, natural language processing, word embeddings

DOI: 10.3233/JIFS-179909

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2497-2507, 2020

Authors: Ángel García-Calderón, Miguel | García-Hernndez, RenArnulfo | Ledeneva, Yulia

Article Type: Research Article

Abstract: There is a lot of cultural heritage information in historical documents that have not been explored or exploited yet. Lower-Baseline Localization (LBL) is the first step in information retrieval from images of manuscripts where groups of handwritten text lines representing a message are identified. An LBL method is described depending on how the features of the writing style of an author are treated: the character shape and size, gap between characters and between lines, the shape of ascendant and descendant strokes, character body, space between characters, words and columns, and touching and overlapping lines. For example, most of the supervised …LBL methods only analyze the gap between characters as part of the preprocessing phase of the document and the rest of features of the writing style of the author are left for the learning phase of the classifier. For such reason, supervised LBL methods tend to learn particular styles and collections. This paper presents an unsupervised LBL method that explicit analyses all the features of the writing style of the author and processes the document by windows. In this sense, the proposed method is more independent from the writing style of the author, and it is more reliable with new collections in real scenarios. According to the experimentation, the proposed method surpasses the state-of-the-art methods with the standard READ-BAD historical collection with 2,036 manuscripts and 132,124 manually annotated baselines from 9 libraries in 500 years. Show more

Keywords: Lower-baseline localization, historical document analysis, text line segmentation, writing style features

DOI: 10.3233/JIFS-179910

Citation: Journal of Intelligent & Fuzzy Systems, vol. 39, no. 2, pp. 2509-2520, 2020