Textual Statistics Research Articles

Semi-Supervised Distributional Vector Generation Techniques for Text Classification

Text class has loved its privilege as a core studies area in text mining. Supervised, unsupervised are the 2 famous paradigms within the technique of type. Relatively novel method of classification is semi-supervised mastering which is midway among the supervised and unsupervised getting to know. With smaller schooling statistics units and taking the large without problems to be had unlabeled data, the procedure of studying in class is refined. There are versions in semisupervised, transductive gaining knowledge of wherein the trained and untrained facts are given in advance the classifier is built, the goal is to expect the magnificence label of untrained data. The opposite version is inductive learning in which the labeled and unlabeled statistics is utilized in model constructing; goal of the version is to predict the unseen information magnificence label. The paper aims to using transductive getting to know to classifying the textual statistics with the aid of considering the phrases appearing in special parts of the record. The words performing inside the introductory and conclusion a part of the files may additionally play important function within the procedure of type, than the ones seemed in other parts. The approach employed could provide one of a kind weights to words primarily based on their presence in one-of-a-kind role of the document. Taking into consideration the above within the procedure of mapping the textual facts into numerical patterns editions of distributed vector generations are acquired. Taking into account large differences in the duration of the documents, distinct normalization techniques are employed which gave eights one-of-a-kind vectors. Non-parametric, most effective to put into effect ok-nearest neighbour algorithm is hired for free-go with the flow textual classification. The outcomes received conclude that semi-supervised textual class can be carried out without loss in category accuracy where restrained skilled records is to be had, as the accuracies of the gaining knowledge of model in supervised and emi-supervised coincide with each other.

Perceptions of consumers in the financial industry under a qualitative data analysis methodology

This research is developed in order to better understand the feelings, sentiments, experiences and perceptions of dissatisfied consumers who use or have used the credit card Capital One in the US market. Capital One Financial Corporation is a US-based bank holding company specialising in credit cards, home loans, auto loans, banking and savings products. In terms of total assets and deposits, Capital One is the eighth largest bank holding company in the USA. This research has been conducted through a total of 1633 written reviews in the website Consumer Affairs applying a qualitative data analysis to discover the key aspects related to consumer's sentiments, feelings, experiences and perceptions. Textual statistics and cumulative coding frequency are applied in order to determine the main concerns of the customers and avoid possible problems in the future from a marketing and organisational perspective.

Women and international assignments

Purpose – Research on female expatriates has been substantially growing over the last decades and particularly in more recent years. Complementing previous thematic analyses of the literature, the purpose of this paper is to apply textual statistics and correspondence analysis to reveal the existing semantic structure of the field of research on female expatriates. Design/methodology/approach – Using correspondence analysis, the authors explored textual data from the abstracts of 151 identified journal articles published in English since 1975. The authors obtained a graphical representation showing the various developmental stages of research on female expatriates. Findings – The authors found that research follows a home-host country orientation and advances from an organizational focus toward individual-level studies. The authors identified various directions for future research and especially a strong need for more multilevel approaches to study men’s and women’s expatriate experiences and trajectories in various contexts. Research limitations/implications – Only articles with abstracts entered the analysis, which in turn was dependent on the content and quality of these abstracts. This limitation has been addressed by thoroughly reading each article considered. Originality/value – This review adopts an original method in research on (female) expatriates and more broadly management research. It enabled the authors to map out the development of key research themes over time. Based on this analysis, gaps in current research could be identified and clear directions for future research were formulated.

Rhetorical Strategy in Forensic Speeches: Multidimensional Statistics-Based Methodology

Rhetorical strategy is relevant in the law domain, where language is a vital instrument. Textual statistics have much to offer for uncovering such a strategy. We propose a methodology that starts from a non-structured text; first, the breakpoints are automatically detected and lexically homogeneous parts are identified; then, the shape of the text through the trajectory of these parts and their hierarchical structure are uncovered; finally, the argument flow is tracked along. Several methods are combined. Chronological clustering of multidimensional count series detects the breakpoints; the shape of the text is revealed by applying correspondence analysis to the parts×words table while the progression of the argument is described by labelled time-constrained hierarchical clustering. This methodology is illustrated on a rhetoric forensic application, concretely a closing speech delivered by a prosecutor at Barcelona Criminal Court. This approach could also be useful in politics, communication and professional writing.

Tracking verbal-based methods beyond conventional descriptive analysis in food science bibliography. A statistical approach

The usage of verbal-based methods beyond conventional descriptive analysis is increasing in sensory analysis, either as full-methods or as a complement to holistic methods. They contribute to a better understanding of the consumers’ likes and willingness, important factors to the food industry. A prime objective of this paper is to give a global vision of the scientific publications in food science related to this topic from their abstracts. Textual statistics, combining multidimensional methods such as correspondence analysis, multiple factor analysis for contingency tables and characteristic words, are proving to be useful for extracting information from the corpus of abstracts. These abstracts have evolved over time towards a higher concern for research about methodology, which has become more complex and requires sophisticated statistical methods. Sensory methods, such as free choice profile, flash profile, repertory grid, sorting task, napping, word association and CATA, have emerged or have been revitalised. New statistical methods, such as multiple factor analysis, have been introduced to analyse data issued from verbalisation tasks. However, correspondence analysis, a reference method for dealing with texts and, more generally, frequency tables, is used with too much restraint.

Language/Discourse: Textual Analysis, Computer and Statistics

Abstract: Textual linguistics has been undergoing a continuous redefining of its internal and external boundaries after Halliday & Hasan’s inaugural thesis. It notably concerns the concepts of text and discourse, and has led to the constitution of a new research field: Textual Discourse Analysis (TDA). We will start from two questions which were asked at the origin of this Congress: “What is a text – or the text – in the digital age?” and “Which new methods are available for textual analysis?” We assume that text is neither merely nor principally a linguistic object, but is at the meeting point of various fields and key-concepts: sentence, utterance, document, archive, basis and corpuses, discourse. Answers must be searched in digital philology as well as in the hermeneutic aspects of discourse analysis.

Corpus stylistics in translation studies: two modern Chinese translations of Don Quijote

This article aims to provide a quantitative account of the stylistic differences between two modern Chinese translations of Don Quijote in terms of the use of language archaism. For this purpose, a number of statistical techniques applicable in corpus linguistics have been used, which exemplifies the foregrounding of quantitative primary data. The article provides an original interdisciplinary study which attempts to integrate research methods adapted from different areas of research, e.g. textual statistics, corpus linguistics, literary stylistics and translation studies, into the development of an emerging field known as corpus stylistics.

Un quart de siècle de méthodologie- L'emergence du longitudinal, des réseaux et de la statistique textuelle

A Quarter of a Century of Methodology - The Emergence of Longitudinal Analysis, Social Networks and Textual Statistics: Using the titles, abstracts and key words (in French) of the 318 articles in the BMS, issues 1 to 99, the author analyzes this corpus with hierarchically ascending and descending classification programs and with factoral correspondence analysis. With the results of these three methods, he describes in detail five major thematic classes: Survey Situation, Longitudinal Analysis, Social Network Analysis, Theoretical Debate and Textual Analysis. Their interrelationship and évolution over time is also described.

Unifying Textual and Visual Cues for Content-Based Image Retrieval on the World Wide Web

A system is proposed that combines textual and visual statistics in a single index vector for content-based search of a WWW image database. Textual statistics are captured in vector form using latent semantic indexing based on text in the containing HTML document. Visual statistics are captured in vector form using color and orientation histograms. By using an integrated approach, it becomes possible to take advantage of possible statistical couplings between the content of the document (latent semantic content) and the contents of images (visual statistics). The combined approach allows improved performance in conducting content-based search. Search performance experiments are reported for a database containing 350,000 images collected from the WWW.

Hazardous measures: an interpretive textual analysis of quantitative sensemaking during crises

Summary The current paper presents a computer-supported approach to the interpretive analysis of organizational texts and documents. Computer-supported interpretive textual analysis, as presented here, is a qualitative research approach which seeks to provide insights into members' meanings and interpretations. It uses four analytical processes: theoretical sampling, computer software supported text analysis, expansion analysis, and producing textual statistics. Interpretive textual analysis contrasts with quantitative content analysis methods which compose variables from qualitative data and use these variables to test existent theories using inferential statistics. The present paper uses computer supported textual analysis to investigate quantitative sensemaking during a public inquiry into a well blow-out involving hydrogen sulphide gas. I address three research questions relevant to crisis sensemaking. First, what quantitative practices and terms were used in sensemaking about the crisis, and how were these used? Second, how were quantitative practices relevant to the management of risks and hazards? And third, how did sensemaking vary among stakeholder groups, and what were the implications of the variations for organizational action? The results and findings from the computersupported textual analysis show that quantitative practices and terms played an important role in inquiry sensemaking. The two theoretically meaningful groups involved in the incident used different vocabularies and logics for sensemaking. The government group used a 'step logic' to emphasize formal steps in the management of the hazard. This required the precise measurement of the hazards as a basis for initiatingrule governed actions to control the hazards. In contrast, the operator company used a logic of local safeguarding to interpret the hazards and measurements of the hazards. The paper concludes by discussing the general methodological and theoretical implications which interpretive textual analysis has for studies of sensemaking in organizational behavior research. ? 1997 John Wiley & Sons, Ltd.