Od Informacji do Wiedzy

This study proposes a novel biologically plausible mechanism for generating low-dimensional spike-based text representation. First, we demonstrate how to transform documents into series of spikes (spike trains) which are subsequently used as input in the training process of a spiking neural network (SNN). The network is composed of biologically plausible elements, and trained according to the unsupervised Hebbian learning rule, Spike-Timing-Dependent Plasticity (STDP). After training, the SNN can be used to generate low-dimensional spike-based text representation suitable for text/document classification. Empirical results demonstrate that the generated text representation may be effectively used in text classification leading to an accuracy of 80.19% on the bydate version of the 20 newsgroups data set, which is a leading result amongst approaches that rely on low-dimensional text representations.

In this paper, the author presents a novel information extraction system that analyses fire service reports. Although the reports contain valuable information concerning fire and rescue incidents, the narrative information in these reports has received little attention as a source of data. This is because of the challenges associated with processing these data and making sense of the contents through the use of machines. Therefore, a new issue has emerged: How can we bring to light valuable information from the narrative portions of reports that currently escape the attention of analysts? The idea of information extraction and the relevant system for analysing data that lies outside existing hierarchical coding schemes can be challenging for researchers and practitioners. Furthermore, comprehensive discussion and propositions of such systems in rescue service areas are insufficient. Therefore, the author comprehensively and systematically describes the ways in which information extraction systems transform unstructured text data from fire reports into structured forms. Each step of the process has been verified and evaluated on real cases, including data collected from the Polish Fire Service. The realisation of the system has illustrated that we must analyse not only text data from the reports but also consider the data acquisition process. Consequently, we can create suitable analytical requirements. Moreover, the quantitative analysis and experimental results verify that we can (1) obtain good results of the text segmentation (F-measure 95.5%) and classification processes (F-measure 90%) and (2) implement the information extraction process and perform useful analysis.

In this paper, the author presents a novel information extraction system that analyses fire service reports. Although the reports contain valuable information concerning fire and rescue incidents, the narrative information in these reports has received little attention as a source of data. This is because of the challenges associated with processing these data and making sense of the contents through the use of machines. Therefore, a new issue has emerged: How can we bring to light valuable information from the narrative portions of reports that currently escape the attention of analysts? The idea of information extraction and the relevant system for analysing data that lies outside existing hierarchical coding schemes can be challenging for researchers and practitioners. Furthermore, comprehensive discussion and propositions of such systems in rescue service areas are insufficient. Therefore, the author comprehensively and systematically describes the ways in which information extraction systems transform unstructured text data from fire reports into structured forms. Each step of the process has been verified and evaluated on real cases, including data collected from the Polish Fire Service. The realisation of the system has illustrated that we must analyse not only text data from the reports but also consider the data acquisition process. Consequently, we can create suitable analytical requirements. Moreover, the quantitative analysis and experimental results verify that we can (1) obtain good results of the text segmentation (F-measure 95.5%) and classification processes (F-measure 90%) and (2) implement the information extraction process and perform useful analysis.

This study aims to propose (i) a multi-view text classification method and (ii) a ranking method that allows for selecting the best information fusion layer among many variations. Multi-view document classification is worth a detailed study as it makes it possible to combine different feature sets into yet another view that further improves text classification. For this purpose, we propose a multi-view framework for text classification that is composed of two levels of information fusion. At the first level, classifiers are constructed using different data views, i.e. different vector space models by various machine learning algorithms. At the second level, the information fusion layer uses input information using a features projection method and a meta-classifier modelled by a selected machine learning algorithm. A final decision based on classification results produced by the models positioned at the first layer is reached. Moreover, we propose a ranking method to assess various configurations of the fusion layer. We use heuristics that utilise statistical properties of F-score values calculated for classification results produced at the fusion layer. The information fusion layer of the classification framework and ranking method has been empirically evaluated. For this purpose, we introduce a use case checking whether companies’ domains identify their innovativeness. The results empirically demonstrate that the information fusion layer enhances classification quality. The Friedman’s aligned rank and Wilcoxon signed-rank statistical tests and the effect size support this hypothesis. In addition, the Spearman statistical test carried out for the obtained results demonstrated that the assessment made by the proposed ranking method converges to a well-established method named Hellinger – The Technique for Order Preference by Similarity to Ideal Solution (H-TOPSIS). Thus, the proposed approach may be used for the assessment of classifier performance.

The aim of this study is to provide an overview the state-of-the-art elements of text classification. For this purpose, we first select and investigate the primary and recent studies and objectives in this field. Next, we examine the state-of-the-art elements of text classification. In the following steps, we qualitatively and quantitatively analyse the related works. Herein, we describe six baseline elements of text classification including data collection, data analysis for labelling, feature construction and weighing, feature selection and projection, training of a classification model, and solution evaluation. This study will help readers acquire the necessary information about these elements and their associated techniques. Thus, we believe that this study will assist other researchers and professionals to propose new studies in the field of text classification.

We are building our own application for text processing. It is based on our REST API which integrates four language models. Our API will enable learning and testing. It is simple, compact and ready to use. Our API will help you avoid the time consuming configuration of many language models. Thus you can develop your own solutions and applications.

The OPI Toolkit for NLP is:

• multilingual, i.e. it enables analysis of documents written in different languages: Polish, English, German, and French,
• ready to learning and testing, because it is simple: you can quickly prototype and develop own solution based on our API,
• compact: you can spend time resolving real problems rather than wasting it on configuration and implementation of NLP basic functionalities. Now these features are available immediately.

Check it in Inventorum NLP Tools website

The aim of this study is to propose an information extraction system, called BigGrams, which is able to retrieve relevant and structural information (relevant phrases, keywords) from semi-structural web pages, i.e. HTML documents. For this purpose, a novel semi-supervised wrappers induction algorithm has been developed and embedded in the BigGrams system. The wrappers induction algorithm utilizes a formal concept analysis to induce information extraction patterns. Also, in this article, the author (1) presents the impact of the configuration of the information extraction system components on information extraction results and (2) tests the boosting mode of this system. Based on empirical research, the author established that the proposed taxonomy of seeds and the HTML tags level analysis, with appropriate pre-processing, improve information extraction results. Also, the boosting mode works well when certain requirements are met, i.e. when well-diversified input data are ensured.

The aim of this study was to propose a classification method for documents that include simultaneously text parts in various languages. For this purpose, we constructed a three-leveled classification system. On its first level, a data processing module prepares a suitable vector space model. Next, in the middle tier, a set of monolingual or multilingual classifiers assigns the probabilities of belonging each document or its parts to all possible categories. The models are trained by using Multinomial Naive Bayes and Long Short-Term Memory algorithms. Finally, in the last component, a multilingual decision module assigns a target class to each document. The module is built on a logistic regression classifier, which as the inputs receives probabilities produced by the classifiers. The system has been verified experimentally. According to the reported results, it can be assumed that the proposed system can deal with textual documents which content is composed of many languages at the same time. Therefore, the system can be useful in the automatic organizing of multilingual publications or other documents.

The aim of this study was to describe a found method for detection of logotypes that indicate innovativeness of companies, where the images originate from their Internet domains. For this purpose, we elaborated a system that covers a supervised and heuristic approach to construct a reference dataset for each logotype category that is utilized by the logistic regression classifiers to recognize a logotype category. We proposed the approach that uses one-versus-the-rest learning strategy to learn the logistic regression classification models to recognize the classes of the innovative logotypes. Thanks to this we can detect whether a given company’s Internet domain contains an innovative logotype or not. More- over, we find a way to construct a simple and small dimension of feature space that is utilized by the image recognition process. The proposed feature space of logotype classification models is based on the weights of images similarity and the textual data of the images that are received from HTMLs ALT tags.

The objective of this paper was to propose a classification method of innovative domains on the Internet. The proposed approach helped to estimate whether companies are innovative or not through analyzing their web pages. A Naïve Bayes classification committee was used as the classification system of the domains. The classifiers in the committee were based concurrently on Bernoulli and Multinomial feature distribution models, which were selected depending on the diversity of input data. Moreover, the information retrieval procedures were applied to find such documents in domains that most likely indicate innovativeness. The proposed methods have been verified experimentally. The results have shown that the diversified classification committee combined with the information retrieval approach in the preprocessing phase boosts the classification quality of domains that may represent innovative companies. This approach may be applied to other classification tasks.