COMPARATIVE ANALYSIS OF METHODS OF VECTORIZATION OF HIGH DIMENSIONAL TEXT DATA

The presented publication is devoted to an overview of the problem presenting textual informationfor subsequent implementation cluster analysis in framework processingand managing high-dimensional information. Modern requirements for analytical, search andrecommendation information systems demonstrate weak formation a holistic solution thatcan provide sufficient level speed and quality results obtained within ofthe current technology market. problementails need conduct objective existing solutions representing informationin vector space, order form view advantages disadvantages analyzed approaches, as well criteria that allow one implement theirown approach, devoid identified weaknesses. work allows youto get idea state elaboration limited subjectarea. Clustering text data automatic subsets, elements which are instancesof documents some researched, unstructured sample fixed dimension. This processcan be classified unsupervised learning, implies absence expert who personallyassigns class indices original documents. However, clusteranalysis without any pre-processing impossible. To do this, it necessary ensurestandardization reduction input single format form. Within ofthis stage analysis, discusses methodsfor preprocessing data. novelty lies thetheoretical basis main methods vectorization, by systematizing objectifyingthe proposed assumptions, conducting series experimental studies. difference from already published scientific works systematization modernsolutions, hypotheses about relevance effectiveness our own hybridizedapproach designed vectorization.