Mathematical Modeling for Classification of Appointments in a Corporate Calendar Based on Unstructured Data

In the modern management systems, automation of event planning and meeting coordination is beginning to be actively implemented as a time management tool. The author of this work reaches the goal of creating an algorithm for classifying meetings from the corporate calendar of employees of a large company by analyzing unstructured data, as well as other features. In the course of the work, methods of processing text information, such as tokenization, normalization and vectorization, methods of extracting features from text and enriching it with the help of Named Entity Recognition, Latent Semantic Analysis and Latent Dirichlet Allocation were studied and implemented, method of enriching texts with contextual information with the help of Contextual Search Phrases fit for the Russian language was studied and applied, such classification models as the Naive Bayesian Classifier, the Support vector machines and random forest were also studied and applied. As a result, a meeting classification model for the corporate calendar was built. It was successfully implemented in the company's corporate calendar application, where it is used to increase the efficiency of employees' work, as one of the time management tools.