This event is endorsed
and organized by

7th EAI International Conference on Big Data Technologies and Applications

November 17–18, 2016 | Seoul, South Korea
The 1st International Workshop on Digital Humanity with Big Data
(DiHuBiDa 2016)
17-18 November, 2016 – Seoul, South Korea
Affiliated with 7th EAI International Conference on Big Data Technologies and Applications


With emergence of big data, analysis and appliance of unstructured data come to the fore. Above all, contents analysis is being one of the major topics in unstructured data analysis, as the amount of contents distributed on web is exponentially increasing. However the existing contents analysis approaches from metadata-based to affective analysis cannot overcome their inherent limitation which is a semantic gap. It is hard to bridge between low-level semantic features in the contents and high-level semantic features, because bridging them just based on mathematical foundation causes questionable logical leaps.

Accordingly narrative-based contents analysis approaches are being magnified as a complementary method. In the past, analyzing narratives of contents is an own domain of critics who major in the humanities. However, nowadays since a huge amount of contents is published on web, a qualitative analysis for them has no choice but to face with physical limitations. Also the narrative-based contents analysis requires not only knowledge of data mining, pattern recognition, and image processing, but also humanistic background like narrative theory, storytelling, and so on. It makes interdisciplinary study between Humanity and Big Data analysis necessary.

In this workshop, we are focusing on the various issues on narrative-based contents analysis, such as, correlations between human cognition and narratives, analyzing and visualizing narratives, authoring support for narrative contents, mining patterns from narratives, recommender system for narrative contents, and so forth. Also, it is opened to every topics about computational processing of narrative contents.


Marc Cavazza (University of Kent, UK)
Title: The Dual Nature of Narrative Technologies


From a cognitive perspective, narratives have been construed as a privileged medium for transmitting knowledge and experience, although other theories have suggested that they could serve the dual purpose of making sense of our real-world experience. The development of narrative technologies over the past 20 years has primarily sought to recreate meaningful experience by addressing challenges such as action representation and event causality, or controlling the shape of narrative progression. However, the use of narrative technologies to organise user data has received comparatively less attention.
After outlining the fundamentals of narrative generation, this talk will discuss recent progress and trends in narrative technologies, in particular the evolution towards more abstract control of narrative trajectories, which has improved the expressive power of narrative representations and their application beyond entertainment. This aspect will be illustrated through two research prototypes, addressing patient education and biohazard management training.
To introduce how narrative generation can be used for data interpretation, we will examine an experiment in the automatic generation of virtual narratives from the social network describing relationships within the cast of characters, and how these results can be generalised.
The final part of this talk will discuss challenges and opportunities in the use of narrative technologies for data interpretation, considering both relational and temporal aspects of data sets.

Adam Jatowt (Kyoto University, Japan)
Title: Bridging Past with Present: Finding Similar Terms across Time and Analyzing Word Meaning Change


Recently, large amounts of historical documents (e.g., news articles) have been digitized and made accessible to the public. However, we still lack effective methods for accessing and making use of such temporal data. This talk will overview our efforts towards facilitating access and understanding of historical texts. We will first describe the framework for estimating and explaining across-time similarity of terms. For example, given an input term (e.g., iPod) and the target time (e.g. 1980s), the task is to find its counterpart that existed in the target time (e.g., Walkman). The related problem is then to output evidences for helping to understand the term similarity. Our approach relies on transforming word contexts across time based on their neural network representations. In the remaining part of the talk we will showcase online system for exploring semantic changes of words over time.

Pablo Gervás (Universidad Complutense de Madrid., Spain)
Title: Finding Protagonists: are there Stories in Your Data?


People love to get their information in the form of stories, so there is a general trend to try to find narrative in everything. Stories draw our attention because they are about characters, and some characters actually lead our passage through the story. The task of automating the construction of narratives from generic data presents many challenges, including the difficulty of modelling high cognitive faculties of human related to affect, goals and aspirations. But there is a level of narrative structure that concerns characters and how the events are told from the point of view of particular characters. This constitutes a basic mechanism for encoding a cloud of events as a sequence of segments of linear discourse. My talk will address recent efforts to model computationally this type of mechanisms that underlie the composition of narrative discourse. I will review some elementary notions of narratology, and I will present insights arising from work on the simulation of such basic storytelling abilities in a society of agents.