Profiling plays an important role in
personalized recommendation in research social networking
platforms and variety of big data sources. Dynamic changes
of user interests posted in social media and trillions of data
generated through big data sources lead to significant
challenges on comprehensive user profiling. Existing
approaches mostly rely on content-based analysis or itembased
rating mechanisms. Most of these approaches rely on
single sources and largely overlook the power of integration
of multiple data sources especially social media. In light of this
deficiency, this research proposes an integrated novel
approach to automatically capture dynamic user interests.
The proposed method facilitates automated learning of user
profiles based on the proposed approach which integrates
preprocessing and electronic symbol extraction model,
opinion-confidence mining hybrid model, feature extraction
model and profile data aggregation weighted k-means
clustering model and analysis of social data. According to our
experimental analysis the proposed method outperforms
existing benchmark methods.