Social network body constructing method based on machine learning

Abstract

The invention discloses a social network body constructing method based on machine learning. The method is mainly used for determining the hyponymy (namely the inclusion relation on the traditional significance) between tags on a social network and constructing a corresponding body. Tags are captured from the social network to be used as an original data set. The method comprises the steps that firstly, six characteristic values are designed for describing the similarity between any two tags, then, some tag pairs are selected from the original data set, a training data set is constructed with the combination of the similarity characteristic values, and whether the tag pairs in the training data set has the hyponymy or not is labeled artificially; then, a Random Forest machine learning model is utilized for training the training data set to obtain a classifier model; then, the classifier model is utilized for determining and labeling the hyponymy between any two tags in original data; all the tags with the hyponymy are extracted out to construct a final social network body.

Description

technical field [0001] The invention belongs to the field of ontology construction, and relates to a machine learning-based social network ontology construction method. Background technique [0002] In recent years, social networks have developed rapidly, and more and more people have begun to use them. With the popularity of social networks, the amount of data in social networks is also increasing. Many social networking sites allow users to mark and classify some content by customizing tags, which is what we call the focus taxonomy. These classification labels generated through user-defined methods lack standardization, and may have problems such as semantic ambiguity, inaccurate words, polysemous words, and homonyms. This brings great challenges to the construction of social ontology based on segment taxonomy. [0003] There are a large amount of flat, messy, and unorganized data on social networks, which are not fully utilized. Building ontology is a good way to mode...

AI Technical Summary

Problems solved by technology

However, these methods are only based on tag co-occurrence and association rule mining to realize the construction of ontology, and cannot deeply mine the semantic relationship between tags in the ontology, so the construction effect is not satisfactory and does not meet the practical requirements.

Images