Company recommendation model: Empowering the accounting system and publicly available data

Abstract

Business management requires constant decision-making. Usually, the selection of suitable partners for collaboration is very intuitive, experience-based skill or requires the usage of data analytics. The most important aspect to get an explainable recommendation for possible collaboration between companies is data. In this research, a company recommendation model is created to incorporate both the accounting system and publicly available data. The accounting system data is used to estimate the collaboration effectiveness of existing collaboration cases. The publicly available data include company registration data and news portal article texts. The news article data include the detection of the company mentioning, estimation of its sentiment and context category. Separate models were developed for different data extraction and analysis. A combined company recommendation system was designed. Model validation with gathered test cases demonstrated 70% recommendation accuracy.