Meeting of the Association for Computational Linguistics, Jun 19, 2011
Joint sentiment-topic (JST) model was previously proposed to detect sentiment and topic simultane... more Joint sentiment-topic (JST) model was previously proposed to detect sentiment and topic simultaneously from text. The only supervision required by JST model learning is domain-independent polarity word priors. In this paper, we modify the JST model by incorporating word polarity priors through modifying the topic-word Dirichlet priors. We study the polarity-bearing topics extracted by JST and show that by augmenting the original feature space with polarity-bearing topics, the in-domain supervised classifiers learned from augmented feature representation achieve the state-of-the-art performance of 95% on the movie review data and an average of 90% on the multi-domain sentiment dataset. Furthermore, using feature augmentation and selection according to the information gain criteria for cross-domain sentiment classification, our proposed approach performs either better or comparably compared to previous approaches. Nevertheless, our approach is much simpler and does not require difficult parameter tuning.
One of the difficult challenges of any knowledge centric online community is to sustain the momen... more One of the difficult challenges of any knowledge centric online community is to sustain the momentum of knowledge sharing and knowledge creation effort by its members through various means. This requires a clearer understanding of user needs that drive community members to contribute, engage and stay loyal to the community. In this paper, we explore the applicability of Abraham Maslow's theory (1943) to understand user behavior and their latent needs using Exploratory Factor analysis. Results show that users are largely driven by four main needs: social interaction, altruism cognitive need and reputation. Our results further indicate that users with high reputations are more likely to stay longer in the community than others, and that socially motivated users are responsible for increased content creation.
Techniques to rank ontologies are crucial to aid and encourage the re-use of publicly available o... more Techniques to rank ontologies are crucial to aid and encourage the re-use of publicly available ontologies. This paper presents a system that obtains a list of ontologies from a search engine that contain the terms provided by a knowledge engineer and ranks them. The ranking of these ontologies will be done according to how many of the concept labels in those ontologies match a set of terms extracted from a corpus of documents related to the domain of knowledge identified by the knowledge engineer's original search terms.
International Journal of Virtual Communities and Social Networking, Jul 1, 2009
Most organisations store their data in several databases with no flexible mechanism for integrati... more Most organisations store their data in several databases with no flexible mechanism for integration and access, and with no common vocabulary in place. Maintaining local vocabularies while realising distributed access is a challenge that most organisations face regularly. For several years, the Semantic Web community has been developing algorithms for mapping data models (ontologies). Nevertheless, ontology mapping remains to be a great challenge, and humans are always expected to verify the results of existing automatic mapping tools. The spread of social web demonstrate the possibility of using collaborative techniques for reaching consensus and fostering user participation. While a number of prototypes for collaborative ontology construction are being developed, collaborative ontology mapping is not yet well investigated. In this article, the authors describe an approach that combines off-the-shelf ontology mapping tools with social software techniques to enable users to collaborate on mapping ontologies.
Blogs, Wikis, and Social Bookmark Tools have rapidly emerged on the Web. The reasons for their im... more Blogs, Wikis, and Social Bookmark Tools have rapidly emerged on the Web. The reasons for their immediate success are that people are happy to share information, and that these tools provide an infrastructure for doing so without requiring any specific skills. At the moment, there exists no foundational research for these systems, and they provide only very simple structures for organising knowledge. Individual users create their own structures, but these can currently not be exploited for knowledge sharing. The objective of the seminar was to provide theoretical foundations for upcoming Web 2.0 applications and to investigate further applications that go beyond bookmark-and file-sharing. The main research question can be summarized as follows: How will current and emerging resource sharing systems support users to leverage more knowledge and power from the information they share on Web 2.0 applications? Research areas like Semantic Web, Machine Learning, Information Retrieval, Information Extraction, Social Network Analysis, Natural Language Processing, Library and Information Sciences, and Hypermedia Systems have been working for a while on these questions. In the workshop, researchers from these areas came together to assess the state of the art and to set up a road map describing the next steps towards the next generation of social software. Within the last two years, social software on the Web, such as Flickr, Delicious, Bibsonomy, Facebook, etc., has received a tremendous impact with regard to hundred of millions of users. A key factor to the success of social software tools in the Web is their grass-roots approach to sharing of information between users: there are no limitations on the kind of tags users may select. The resulting structures are often called 'folksonomies', that is, 'taxonomies' created by 'folks'. Such systems are also considered to realize a Web version 2.0. The reason is that the initial use of the Web could be characterized by many users consuming what a comparatively small set of producers had developed, whereas with social software on the Web, everyone becomes a prosumer, i.e. someone who produces and consumes content. The
In the world of public misinformation, there are many cases where the information is not false or... more In the world of public misinformation, there are many cases where the information is not false or fabricated, but rather has been manipulated using more subtle techniques such as word replacements, selection of details, omissions and argument distortion. These techniques can have the e↵ect of influencing the reader's frame of mind towards the events reported. We currently lack the necessary tools to uncover such manipulations automatically. In this position paper, we propose an integrated analysis framework and pipeline to identify various narrative signals in news articles; such as structural roles, framing, and subjectivity. By comparing these at the document level and sentence level, it will be possible to highlight di↵erences of narrative techniques used to report the same news events.
De ning exible and consistent methods and artefacts to design for social impact is a current chal... more De ning exible and consistent methods and artefacts to design for social impact is a current challenge for HCI. The ephemeral and vulnerable conditions of people living as refugees add even more questions about the suitability of design methods to the complexity of real -and many times tough -life . In this position paper we brie y introduce two concepts embraced by the Socially-aware Design Approach, the Semiotic Onion and the Basic Block of Culture. We then re ect about the potential contributions of applying these concepts and artefacts to inform design for boosting community resilience of people living as refugees.
The focus of the AI, The Fundamental Social Aggregation Challenge, and the Autonomy of Hybrid Age... more The focus of the AI, The Fundamental Social Aggregation Challenge, and the Autonomy of Hybrid Agent Groups symposium was to explore issues associated with the control of teams of humans, autonomous machines, and robots working together as hybrid agent groups. Bill Lawless of Paine College kicked off the meeting by pointing out the need for a new theory of social dynamics. He showed that majority rule is far better than consensus for group decision processes and proposed a new mathematical model for characterizing social group dynamics based on interdependence. Albert-Lazlo Barabasi of Northeastern University showed how scale-free networks are very common, and that they are robust to random failures but susceptible to targeted attacks. The hubs in social networks are often not the managers or people in charge. He then explored how to control complex networks. He showed that spare and heterogeneous networks are harder to control than homogeneous networks. Jonathan Barzilai of Dalhousie University pointed out that the current prevailing mathematical foundations of the social sciences are in error because ordinal elements are not vectors, cannot be added or multiplied, do not live in vector spaces, and cannot be differentiated.
Multivariant Information Management and Exploitation - Scenario Specification
This report provides a piloting scenario for MIMEX development which involves UK military operati... more This report provides a piloting scenario for MIMEX development which involves UK military operations in the Helmand province, Afghanistan. The scenario aims to highlight some of the challenges and opportunities for information exploitation in support of stabilising efforts un-dertaken in a complex environment where there is a blur borderline separating enemy forces and civil elements. The scenario describes the situation in Helmand and UK military tasks planned to be carried out in the province for the following 6 months. Aiming at providing sustainable security for the region, the tasks range from ensuring the safety for the local populations to suppressing insurgent activities. All these are information intensive tasks and draws attention to the challenges with respect to effective information acquisition, management and exploitation. Against this scenario, the report analyses the information requirements for the scenario tasks, which encompass both military and civil information. Based on these, it provides a taxonomy of the required information to serve as the departure point for later ontology development work and identifies relevant information resources currently available in the public domain. These resources are potential information sources for MIMEX’s later information acquisition work. In addition to the above, the report also outlines a set of capabilities anticipated to be imple-mented in support of the following areas in the context of the scenario: knowledge base main-tenance, knowledge monitoring (for situation awareness), situation assessment and analysis, and operational planning. These capabilities are presented as a vision of the technologies that MIMEX will develop and to showcase a wide range of benefits that advance knowledge management and processing can bring to military users. Finally, the report identifies and reviews a number of existing semantic technologies that are relevant to the development of MIMEX.
Joint Proceedings of the 1st Workshop on Semantic Sentiment Analysis (SSA2014), and the Workshop on Social Media and Linked Data for Emergency Response (SMILE 2014) co-located with 11th European Seman
Uploads
Papers by Harith Alani