This study presents a new implicit user interest model that aims to improve the accuracy of implicit interest identification. The proposed model is based on previous research [16] that has shown that the relatively low accuracy of implicit interest identification is often due to the fact that the predicted implicit interests are not actually the user's true interests.
To address this issue, our proposed model includes three key components: task sentiment analysis, expanded entities (i.e., interest candidates), and entity scoring. By incorporating task sentiment analysis, our model is able to take into account the sentiment of the user's explicit interests, which has been shown to be an important factor in identifying implicit interests [16]. By expanding the pool of entities considered as potential interests, our model is able to consider a broader range of potential interests, increasing the chances of identifying the user's true interests. Finally, by scoring entities based on their relevance and importance, our model is able to prioritize the most relevant and important interests, further improving the accuracy of implicit interest identification. Our proposed implicit user interest model aims to produce more accurate results than previous studies, and has the potential to significantly improve the effectiveness of implicit interest identification systems.