Open source intelligence for cyber situational awareness related to concepts, models and methods for cyber threat intelligence in tactical, operational and strategic levels are central
to the scope of this event. Cyberattacks are becoming increasingly complex and routine threats emerging from
an evolving threat landscape. Security breaches frequently compromise sensitive information, exposing personal
identities, intellectual property or financial assets. Things will get a lot worse when advanced attacks target
public infrastructure on which we all depend. Besides temporal disruption of critical services, sophisticated attacks can cause physical damage to vital systems like the electric grid or
intelligent transportation systems.
Criminal networks and terrorist groups often operate globally, hiding their illicit activities by taking advantage of information and communications technology. Digital communication enables global communities that are hard to track. Open source intelligence can be a powerful tool for combating crime by linking global and local crime patterns to help understand how criminal activities are connected across crime spaces and emerging trends such as the inevitable shift of conventional organized crime to the cyber arena.
Cyber threat information can be drawn from countless sources from the dark and surface web such as hacker forums, darknet markets, internet relay chats, and social media. This situation calls for radical advances in using cutting-edge data mining, machine learning and predictive analytics combined with social network analysis and natural language processing to detect, disrupt and neutralize cyber and physical threats. We invite submissions of original research papers describing solutions and systems for all aspects of open source intelligence and security informatics. Topics of interest include but are not limited to cyber threat intelligence, cybercrime analysis, web intelligence and security, big crime data mining algorithms and open source situational awareness.
|Full paper submission deadline||May 01, 2019 11:59 PM American Samoa Zone (UTC-11)|
|Notification of acceptance||June 25, 2019|
|Camera-ready papers due||July 15, 2019 11:59 PM American Samoa Zone (UTC-11)|
|Conference events||August 27-30, 2019|
Because most of language production is outside conscious control,
it provides a channel that is of considerable interest in intelligence contexts.
Analytics applied to natural language operates in two modes. The first might be
called reverse engineering and tries to infer properties of speakers and writers
such as their attitudes, personality, affective state (moods and emotions),
and mental health. The second looks how language is being leveraged,
both consciously and unconsciously, for purposes such as influence,
propaganda, and abuse. There is some overlap between these two modes.
For example, understanding the language production mechanisms of abusive
language helps us detect and block it, but also understand the
mental state of those who produce it, and their tactics for using it.
From a technical perspective, there are three approaches that have been successful. Deep learning models such as biLSTMs can detect most of these properties with high accuracy, but they are black boxes, and so we learn little about cognitive and linguistic Predictive techniques based on bag-of-word data tend to be less accurate but, with attribute ranking, make it possible to elicit the connection between language and effects. Finally, hybrid techniques are being developed using word embeddings, in the style of FastText, as if they were words. Clustering words based on their embedding encodings has two advantages: document-word matrices can be biclustered, and word synonyms can be discovered even when they are the one-off made-up words common in online text. I will illustrate how these ideas work out in practice in real-world data, and how this can be useful in intelligence.
We welcome all submissions reporting original and unpublished research written in English and pertaining to the topics mentioned in the Call For Paper page. Three categories of submissions will be accepted: full papers up to 8 pages, short communications up to 4 pages, and abstracts up to 2 pages, all in IEEE two-column format. Submissions should include the title, authors, affiliation, e-mail address, tel/fax numbers, abstract, and postal address on the first page. Papers should be submitted online using the symposium Online Submission System. Paper submission implies one of the authors must register and attend the symposium to present the paper. FOSINT-SI 2019 participants will also enjoy free access to the full program of ASONAM 2019 and all the accompanying symposia and workshops.
Papers will be thoroughly reviewed by experts in the field and will be selected based on their originality, timeliness, significance, relevance, and clarity of presentation. Further, authors should certify that their submissions contain substantially new and previously unpublished research. Accepted and presented papers will be included in the FOSINT-SI 2019 Conference Proceedings and forwarded for inclusion in IEEE Computer Society Digital Library (CSDL), IEEE Xplore and the ACM Digital Library. The conference proceedings will be submitted for EI indexing through INSPEC by IEEE. The proceedings will be also covered by several other indexes, including DBLP, SCOPUS, etc. A selection of high-quality research papers from FOSINT-SI 2019 will be invited to submit an extended version for an edited book in the Lecture Notes in Social Networks (LNSN) series published by Springer.