The 2nd International Workshop on Spatial/Temporal Information Extraction from Unstructured Texts (WSTIE 2017)
May 29, 2017, Daejeon, South Korea
In conjunction with the IEEE MDM 2017 - 18th IEEE International Conference on Mobile Data Management
As the number of unstructured documents grows exponentially, it becomes important to develop methods for automatically extracting information from the documents. Especially, the documents usually have temporal information (e.g., today, June 2000) and spatial information (e.g., south from here, Hongkong), so it is necessary to develop the methods for extracting such useful information from the texts. The spatial/temporal information can be used for various applications, such as Question-Answering (QA) systems, Knowledge-Base (KB) construction, Recommendation systems, and Context-Aware services on mobile platforms. Developing methods for extracting spatial/temporal information involves several challenging issues: we need to (1) deal with the exponentially growing number of documents (i.e., big data), and (2) investigate language-specific issues for achieving high performances. We want to discuss and share the knowledge about how to solve these issues, and look for a way to develop better methods for extracting the information.
This workshop aims to create opportunities to discuss about the state-of-the-art studies, and to share on-going works. We hope that this will enhance collaboration among the researchers related to spatial/temporal information extraction. Although there is a well-known shared-task series, namely TempEval, we see that it has a limitation that it does not consider deeply language-specific issues which must be solved for achieving high performances. Thus, we want to foster a new chance for the discussion about not only the state-of-the-art studies, but also some language-specific issues. We also hope to have opportunity to discuss about how to combine the spatial information with the temporal information. As these two types of information are strongly related to each other, we believe that it must be better to consider both of them at a time, rather than dealing with them independently. Finally, we aim to discuss about how the spatial/temporal information can be helpful for various context-aware applications on mobile platforms.
We invite submissions on topics that include, but are not limited to, the following:
Prospective authors are invited to submit their papers, 4~6 pages, in English according to the IEEE two-column format for conference proceedings. The direct link for paper submission is https://easychair.org/conferences/?conf=wstie2017. All submissions will be peer-reviewed by the Program Committee of the workshop. All accepted workshop papers will be published through IEEE Computer Society. See templates here.
There will be one best paper award with a special prize, Leopold keyboard "fc900r". We hope the award with the special prize will encourage the students/researchers to make a great contribution to the field!
When providing a service based on points-of-interest (POIs), it is imperative to understand the properties and characteristics of the POIs. We present our ongoing project to extract POI contexts from unstructured blog data. Because the data are geo-tagged, we are able to focus on contexts that describe the space or time at which the blogger visited the POI. Examples of extracted contexts include cherry blossom, fountain, royal palace, evening, weekend, summer, Valentine's Day. In addition to temporal and spatial contexts, we extract other contextual factors that are related to the user's experience at the POI, such as quiet, romantic, relaxing, traditional Korean style, etc. Our model uses CNN and LSTM networks to extract POI contexts, which are then used in various different services.
Ahn Seminar #4420, CS Dept (E3-1), KAIST