A New Multifractal-based Deep Learning Model for Text Mining

Text mining aims to automatically and efficiently uncover and explore valuable information or patterns from noisy, irregular and unstructured texts [1-2], thereby enabling us to gain a deeper understanding of the underlying meaning and context within the text, and easily exploring the knowledge, uncovering hidden insights. It can generate informed understanding of the content and has become significant in decision-making in various sectors and domains across industries. For example, we can understand users' preferences [3], sentiments [4], opinions [5], concerns [6] and 2 interests [7] etc., by mining the text generated by users, thus infer their intentions and purposes [8-11]. We are also amenable to the attainment of more sophisticated security risk management practices [12]. Additionally, text mining is responsible for various natural language processing applications such as knowledge graph [13-14], questionanswer dialogue system [15-16], and recommendation system [17-18]. Text mining mainly approaches entity recognition and text classification, both of which exhibit certain distinctions in their form. The purpose of entity recognition is to automatically identify expected knowledge from text [19], such as defect knowledge and technical terms in technical reports [20-21].