Non-structural data acquisition technology based on machine learning

The invention discloses a non-structural data collection technology based on machine learning, which comprises a non-structural data collection general library, preprocessing, feature engineering, a machine learning algorithm and a data application library, and is characterized in that the non-structural data collection general library is arranged to collect various types of acquired data in real time, and preprocessing and feature engineering are carried out to obtain a data application library; impurity quality information in big data is cleaned, meaningful data with computable features can be conveniently extracted from original unstructured data, a machine learning algorithm is applied to perform analysis tasks on the data, the unstructured data can be efficiently analyzed on the basis of a machine learning-based method, and the analysis efficiency is improved. By arranging the data application library, a user can flexibly use the collected data by using result-oriented search at any time, so that the dat