Predicting and Recommending the next Smartphone Apps based on Recurrent Neural Network

Abstract

The popularity of smartphones has witnessed the rapid growth of the number of mobile applications. Nowadays, there are millions of applications available, and at the same time, many applications are already installed on people's smartphones. Installing numerous apps will cause some troubles in finding the specific apps promptly. Hence it is necessary to predict the next app(s) to be used in a short term and launching them as shortcuts, which will make the smartphone system more efficient and user-friendly. In this paper, we pay attention to two subproblems that are related to the app usage prediction. One is the T app prediction problem that focuses on predicting a set of apps that will be used in a time interval. The other is the Top-K app recommendation problem that focuses on recommending the K most probable APPs to be used next. In order to solve these problems, we propose a generic prediction model based on Long Short-term Memory (LSTM), which is an enhancement of the recurrent neural network (RNN) model. The proposed model converts the temporal-sequence dependency and contextual information into a unified feature representation for next app prediction. We implement the model in the Android platform. Extensive experiments based on real collected dataset demonstrate that the proposed LSTM model outperforms the baselines for app usage prediction, and achieves high accuracy for app recommendation.

What and How long: Prediction of Mobile App Engagement

“…After 2017, neural approaches have become increasingly popular. Some researchers proposed different neural models for predicting the next app, including CNN [46] and LSTM [61], etc. Xu et al [61] proposed a generic prediction model based on Long Short-term Memory (LSTM), to covert the temporal-sequence dependency and contextual information into a unified feature representation for next app prediction and stated that it outperforms other models.…”

“…Some researchers proposed different neural models for predicting the next app, including CNN [46] and LSTM [61], etc. Xu et al [61] proposed a generic prediction model based on Long Short-term Memory (LSTM), to covert the temporal-sequence dependency and contextual information into a unified feature representation for next app prediction and stated that it outperforms other models. Additionally, Tian et al [54] explored the prediction to identify if the pair of two app usage logs belong to the same task.…”

“…In summary, for the traditional next app prediction task, we selected four models from previous works that could be fitted to our dataset as baselines, include three works [49,51,70] based on the temporal pattern and contextual features and one recent work based on neural approach (LSTM) [61]. For the joint prediction problem of predicting the next app and engagement level together, since there is no existing similar joint prediction work (app dwell time has a high dependency on which app user is engaging), we added those two recent works [57,59] for predicting dwell time as baselines.…”

“…We then tested additional three methodologies from previous works which conducted the next app prediction by combining app usage history and contexts in a unified manner: SVM+Context [49], CPD [51] and BN [70]. Recently, the neural approaches are popular in solving the time sequence problems, we also investigated the performance of app usage prediction based on LSTM [61] as the baseline:…”

“…Next app prediction is characterised as the willingness to use an app, whereas engagement is the usage pattern after accessing the app. Though most researchers have focused on measuring which app user will use [2,34,35,61], the engagement of apps is still not well understood. Hence, we consider which app user will use and the engagement level of how long the user will stay with this app an aggregated measure of users' app usage behaviour.…”