ABSTRACT. To accomplish the recommendation task of predicting the next item in the scenario that user profile and past activities are limited, session-based recommendation is proposed to model anonymous click sequences. In order to consider complex transitions of items and the correlative information of neighboring sessions simultaneously, an end-to-end Multihop Inter-session Recommendation Framework (MIRF) is proposed in this paper. Specifically, each session sequence is modeled as a directed graph to reveal complex transitions firstly, after which two parallel modules are incorporated: the Intra-session Encoder (Intra-SE) and the Inter-session Encoder (Inter-SE). The Intra-SE models users’ global and current preference via attention network, and the Inter-SE makes better use of inter-session correlation by investigating neighboring sessions. A multihop gated memory network is employed to integrate the outputs of these two encoders and then we obtain the final session representation. Extensive experiments conducted on two real-world datasets demonstrate the effectiveness of MIRF over state-of-the-art baselines.

ABSTRACT. Opinion mining from mobile app reviews has grown exponentially during the last decade. Most studies in this area, however, have focused on a sentiment analysis. In this study, we consider review mining from another perspective, that is, capturing user justifications behind whatever actions are explicitly stated in their app reviews, e.g., the reason behind user purchases. This study highlights how different app features can promote different user decisions, which in turn can be beneficial for software developers to gain valuable data-driven requirements for the planning and development of application updates. We collected, used, and shared our manually annotated 46k mobile app reviews from 12 different app categories in the Google Play Store. We designed three classification problems to filter reviews containing both arguments and decisions from non-argumentative reviews. We extracted three features, namely, structural, lexical, and contextual, from the body of review sentences. Four classifiers (naive Bayes, logistic regression, support vector machine, and random forest) and different feature combinations were trained on the dataset and evaluated to examine if such features can allow us to classify user arguments and decisions. The results show an improved performance over previous studies and show the efficacy of the proposed approach compared to human-assessments.

ABSTRACT. Designing an Intelligent Tutoring System (ITS) that simulates human learning with regard to different knowledge levels is a challenge as it reflects an accurate way of estimating the students’ performance level. Most developed ITSs typically focus on the normal cognitive factors such as the students’ prior success and failure scores without paying appropriate consideration to the sensitive cognitive factors that have a great impact on the student performance prediction such as the integration of the human current skills and given items skills, particularly when the learning items require multiple skills, which thus reduce student’s learning efficiency due to an incomplete representation of the student’s knowledge. This paper presents a modified student modeling approach, called modified Performance Factor Analysis (ModPFA), based on a previously developed model called Performance Factor Analyses (PFA) . ModPFA was developed by adding the hinting parameter to the original PFA formula. This extension has scoring procedure and knowledge level estimation for each student. Results have shown great improvement in terms of performance estimation in the ModPFA compared to PFA.