Semantic relationships between words provide relevant information about the whole idea in the texts. Existing embedding representation models characterize each word as a vector of numbers with a fixed length. These models have been used in tasks involving text classification, such as recommendation and question–answer systems. However, the embedded information provided by semantic relationships has been neglected. Therefore, this paper proposes an approach that involves semantic relationships in embedding models for text classification, which is evaluated. Three embedding models based on semantic relations extracted from Wikipedia are presented and compared with existing word-based models. Our approach considers the following relationships: synonymy, hyponymy, and hyperonymy. They were considered since previous experiments have shown that they provide semantic knowledge. The relationships are extracted from Wikipedia using lexical-syntactic patterns identified in the literature. The extracted relationships are embedded as a vector: synonymy, hyponymy–hyperonymy, and a combination of all relationships. A Convolutional Neural Network using semantic relationship embeddings was trained for text classification. An evaluation was carried out for the proposed relationship embedding configurations and existing word-based models to compare them based on two corpora. The results were obtained with the metrics of precision, accuracy, recall, and $F_1$-measure. The best results for the 20-Newsgroup corpus were obtained with the hyponymy–hyperonymy embeddings, achieving an accuracy of 0.79. For the Reuters corpus, $F_1$-measure and recall of 0.87 were obtained using synonymy–hyponymy–hyperonymy.

Topic discovery involves identifying the main ideas within large volumes of textual data. It indicates recurring topics in documents, providing an overview of the text. Current topic discovery models receive the text, with or without pre-processing, including stop word removal, text cleaning, and normalization (lowercase conversion). A topic discovery process that receives general domain text with or without processing generates general topics. General topics do not offer detailed overviews of the input text, and manual text categorization is tedious and time-consuming. Extracting topics from text with an automatic classification task is necessary to generate specific topics enriched with top words that maintain semantic relationships among them. Therefore, this paper presents an approach that integrates text classification for topic discovery from large amounts of English textual data, such as 20-Newsgroups and Reuters Corpora. We rely on integrating automatic text classification before the topic discovery process to obtain specific topics for each class with relevant semantic relationships between top words. Text classification performs a word analysis that makes up a document to decide what class or category to identify; then, the proposed integration provides latent and specific topics depicted by top words with high coherence from each obtained class. Text classification accomplishes this with a convolutional neural network (CNN), incorporating an embedding model based on semantic relationships. Topic discovery over categorized text is realized with latent Dirichlet analysis (LDA), probabilistic latent semantic analysis (PLSA), and latent semantic analysis (LSA) algorithms. An evaluation process for topic discovery over categorized text was performed based on the normalized topic coherence metric. The 20-Newsgroups corpus was classified, and twenty topics with the ten top words were identified for each class. The normalized topic coherence obtained was 0.1723 with LDA, 0.1622 with LSA, and 0.1716 with PLSA. The Reuters Corpus was also classified, and twenty and fifty topics were identified. A normalized topic coherence of 0.1441 was achieved when applying the LDA algorithm, obtaining 20 topics for each class; with LSA, the coherence was 0.1360, and with PLSA, it was 0.1436.

Information Technologies have generated large amounts of documents available for analysis and use. Information systems can provide the user with the necessary data for a specific purpose without human intervention, saving time in providing the response expected by the user. Some traditional models of topic discovery provide essential information in the literature, but it is still necessary to incorporate the knowledge that a person can use when reading a document. In this work, an analysis of the behavior of the techniques of Latent Dirichlet Analysis, Latent Semantic Analysis, and Probabilistic Latent Semantic Analysis is carried out incorporating the semantic relationships of the type hypernym, hyponym, synonymy, holonymy, and meronymy extracted from an external source of knowledge as WordNet. In order to improve the results obtained by applying the three mentioned techniques in a set of documents without adding external knowledge. Compared to the initial results, our experimental results improved when incorporating semantic relationships, such as hypernyms and synonyms. The best result was obtained when using the Lesk algorithm for word sense disambiguation and subsequently applying Latent Dirichlet Analysis.

Embedding representation models characterize each word as a vector of numbers with a fixed length. These models have been used in tasks involving text classification, such as recommen- dation and question-answer systems. Semantic relationships are words with a relationship between them providing a complete idea to a text. Therefore, it is hypothesized that an embedding model involving semantic relationships will provide better performance for tasks that use them. This paper presents three embedding models based on semantic relations extracted fromWikipedia to classify texts. The synonym, hyponym, and hyperonym semantic relationships were the ones considered in this work since previous experiments have shown that they are the ones that provide the most semantic knowledge. Lexical-syntactic patterns present in the literature were implemented and subsequently applied to the Wikipedia corpus to obtain the semantic relationships present in it. Several semantic relationships are used in different models: synonymy, hyponym-hyperonym, and a combination of the first two. A convolutional neural network was trained for text classification to evaluate the performance of each model. The results obtained were evaluated with the metrics of precision, accuracy, recall, and F1-measure. The best values obtained with the second model were accuracy of 0.79 for the 20-Newsgroup corpus. F1-measure and recall of 0.87 respectively for the Reuters corpus.