Design and evaluation of an English speaking recommender system using word networks and context-aware techniques

Abstract

Speaking is an important part of language communication, but mastering speaking skills and rich vocabulary is a challenge for non-native speakers. To address this problem, the study designs an English speaking recommendation system that incorporates word networks as well as user-based context-aware and semantic-aware techniques. The word network identifies associations between words by analyzing extensive English-speaking learning resources. While the user-based context-aware and semantic-aware techniques generate recommendation results by analyzing the similarity of users. Meanwhile, Precision, Recall, and F1 were selected for comparison. Precision indicates the fraction of cases labeled as positive that are truly positive. Recall indicates the fraction of actual positive cases that are correctly predicted as positive. The F1 score is the average of precision and recall, calculated as their harmonic mean. The results show that each recommendation evaluation index of the spoken English recommendation system shows a tendency of increasing and then decreasing with the increase of the interest similarity weight value W_ic. Meanwhile, the individual recommendation evaluation indexes of the English speaking recommender system decrease with the increase of α value. Among them, when the α value is 0.5, the system has the best recommendation results with Precision, Recall, and F1 as high as 75.69 %, 25.81 %, and 37.48 %, respectively. In addition, each interface of the system provides good response time under high concurrency. In particular, the response time of the login interface is as low as 120 ms and the response time of the recommendation interface is as high as 328 ms on average. Compared to current recommendation systems, the average response time has been improved by at least 50 ms. It shows that the English speaking recommendation system designed by the Institute has good performance advantages and feasibility. It provides reliable technical support for modern spoken English teaching.