Pre-training Context and Time Aware Location Embeddings from Spatial-Temporal Trajectories for User Next Location Prediction

نویسندگان

چکیده

Pre-training location embeddings from spatial-temporal trajectories is a fundamental procedure and very beneficial for user next prediction. In the real world, usually has variable functionalities under different contextual environments. If exact functions of in trajectory can be reflected its embedding, accuracy prediction should improved. Yet, existing pre-trained on are mostly based distributed word representations, which mix location's various into one latent representation vector. To address this problem, we propose Context Time aware Location Embedding (CTLE) model, calculates vector with consideration specific neighbors trajectories. way, multi-functional properties locations properly tackled. Furthermore, order to incorporate temporal information embeddings, subtle encoding module novel pre-training objective, further improve quality embeddings. We evaluate our proposed model two real-world mobile datasets. The experimental results demonstrate that, compared embedding methods, CTLE pre-train higher significantly performance downstream models.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Unsupervised Spatio-Temporal Embeddings for User and Location Modelling

Many social network applications depend on robust representations of spatio-temporal data. In this work, we present an embedding model based on feed-forward neural networks which transforms social media check-ins into dense feature vectors encoding geographic, temporal, and functional aspects for modelling places, neighborhoods, and users. On the basis of this model, we further propose a Spatio...

متن کامل

Mobile Location Prediction in Spatio-Temporal Context

The increasing use of mobile devices and popular mobile services has led to massive availability of mobile data. Location prediction is a specific topic in mobile data mining, with its potential application in traffic planning, location-base advertisement, and user oriented coupon dispersion. Traditional location prediction methods often separately consider spatial or temporal approach. Althoug...

متن کامل

PhotoMap: from location and time to context-aware photo annotations

Despite the growth of geotagged multimedia data on the Web, spatial and temporal metadata are still poorly exploited by Web search engines and multimedia systems. Interpretation and inference processes can enlarge these metadata towards high quality information that are useful for annotating multimedia automatically. In this paper, we propose a semi-automatic approach for annotating personal ph...

متن کامل

What Do Context Aware Electronic Alerts from Virtual Learning Environments Tell Us about User Time & Location?

The paper describes the analysis of user location and time stamp information automatically logged when students receive and interact with electronic updates from the University’s virtual learning environment. The electronic updates are sent to students’ mobile devices using RSS feeds. The mobile reception of such information can be received in three dimensions of context: time, location and act...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence

سال: 2021

ISSN: ['2159-5399', '2374-3468']

DOI: https://doi.org/10.1609/aaai.v35i5.16548