Automatic generation of effective video summaries

As the popularity of video as an information medium rises, the amount of video content that we produce and archive keeps growing. This creates a demand for shorter representations of videos in order to assist the task of video retrieval. The traditional solution is to let humans watch these videos and write textual summaries based on what they saw. This summarisation process, however, is time-consuming. Moreover, a lot of useful audio-visual information contained in the original video can be lost. Video summarisation aims to turn a full-length video into a more concise version that preserves as much information as possible. The problem of video summarisation is to minimise the trade-off between how concise and how representative a summary is. There are also usability concerns that need to be addressed in a video summarisation scheme. To solve these problems, this research aims to create an automatic video summarisation framework that combines and improves on existing video summarisation techniques, with the focus on practicality and user satisfaction. We also investigate the need for different summarisation strategies in different kinds of videos, for example news, sports, or TV series. Finally, we develop a video summarisation system based on the framework, which is validated by subjective and objective evaluation. The evaluation results shows that the proposed framework is effective for creating video skims, producing high user satisfaction rate and having reasonably low computing requirement. We also demonstrate that the techniques presented in this research can be used for visualising video summaries in the form web pages showing various useful information, both from the video itself and from external sources.

Learning domain ontology for tag recommendation

Recently, user tagging systems have grown in popularity on the web. The tagging process is quite simple for ordinary users, which contributes to its popularity. However, free vocabulary has lack of standardization and semantic ambiguity. It is possible to capture the semantics from user tagging and represent those in a form of ontology, but the application of the learned ontology for recommendation making has not been that flourishing. In this paper we discuss our approach to learn domain ontology from user tagging information and apply the extracted tag ontology in a pilot tag recommendation experiment. The initial result shows that by using the tag ontology to re-rank the recommended tags, the accuracy of the tag recommendation can be improved.

Ontology learning from user tagging for recommendation making

Recently, user tagging systems have grown in popularity on the web. The tagging process is quite simple for ordinary users, which contributes to its popularity. However, free vocabulary has lack of standardization and semantic ambiguity. It is possible to capture the semantics from user tagging into some form of ontology, but the application of the resulted ontology for recommendation making has not been that flourishing. In this paper we discuss our approach to learn domain ontology from user tagging information and apply the extracted tag ontology in a pilot tag recommendation experiment. The initial result shows that by using the tag ontology to re-rank the recommended tags, the accuracy of the tag recommendation can be improved.

Leveraging Web 2.0 data for scalable semi-supervised learning of domain-specific sentiment lexicons

Since manually constructing domain-specific sentiment lexicons is extremely time consuming and it may not even be feasible for domains where linguistic expertise is not available. Research on the automatic construction of domain-specific sentiment lexicons has become a hot topic in recent years. The main contribution of this paper is the illustration of a novel semi-supervised learning method which exploits both term-to-term and document-to-term relations hidden in a corpus for the construction of domain specific sentiment lexicons. More specifically, the proposed two-pass pseudo labeling method combines shallow linguistic parsing and corpusbase statistical learning to make domain-specific sentiment extraction scalable with respect to the sheer volume of opinionated documents archived on the Internet these days. Another novelty of the proposed method is that it can utilize the readily available user-contributed labels of opinionated documents (e.g., the user ratings of product reviews) to bootstrap the performance of sentiment lexicon construction. Our experiments show that the proposed method can generate high quality domain-specific sentiment lexicons as directly assessed by human experts. Moreover, the system generated domain-specific sentiment lexicons can improve polarity prediction tasks at the document level by 2:18% when compared to other well-known baseline methods. Our research opens the door to the development of practical and scalable methods for domain-specific sentiment analysis.

A Virtual Evaluation Track for Cross Language Link Discovery

The Wikipedia has become the most popular online source of encyclopedic information. The English Wikipedia collection, as well as some other languages collections, is extensively linked. However, as a multilingual collection the Wikipedia is only very weakly linked. There are few cross-language links or cross-dialect links (see, for example, Chinese dialects). In order to link the multilingual-Wikipedia as a single collection, automated cross language link discovery systems are needed – systems that identify anchor-texts in one language and targets in another. The evaluation of Link Discovery approaches within the English version of the Wikipedia has been examined in the INEX Link the-Wiki track since 2007, whilst both CLEF and NTCIR emphasized the investigation and the evaluation of cross-language information retrieval. In this position paper we propose a new virtual evaluation track: Cross Language Link Discovery (CLLD). The track will initially examine cross language linking of Wikipedia articles. This virtual track will not be tied to any one forum; instead we hope it can be connected to each of (at least): CLEF, NTCIR, and INEX as it will cover ground currently studied by each. The aim is to establish a virtual evaluation environment supporting continuous assessment and evaluation, and a forum for the exchange of research ideas. It will be free from the difficulties of scheduling and synchronizing groups of collaborating researchers and alleviate the necessity to travel across the globe in order to share knowledge. We aim to electronically publish peer-reviewed publications arising from CLLD in a similar fashion: online, with open access, and without fixed submission deadlines.

A voting mechanism for named entity translation in English-Chinese question answering

In this paper, we describe a voting mechanism for accurate named entity (NE) translation in English–Chinese question answering (QA). This mechanism involves translations from three different sources: machine translation,online encyclopaedia, and web documents. The translation with the highest number of votes is selected. We evaluated this approach using test collection, topics and assessment results from the NTCIR-8 evaluation forum. This mechanism achieved 95% accuracy in NEs translation and 0.3756 MAP in English–Chinese cross-lingual information retrieval of QA.

A longitudinal study of corporate earnings guidance in Australia’s continuous disclosure environment

Since the introduction of a statutory‐backed continuous disclosure regime (CDR) in 1994, regulatory reforms have significantly increased litigation risk in Australia for failure to disclose material information or for false and misleading disclosure. However, there is almost no empirical research on the impact of the reforms on corporate disclosure behaviour. Motivated by the absence of research and using management earnings forecasts (MEFs) as a disclosure proxy, this study examines (1) why managers issue earnings forecasts, (2) what firm‐specific factors influence MEF characteristics, and (3) how MEF behaviour changes as litigation risk increases. Based on theories in information economics, a theoretical framework for MEF behaviour is formulated which includes antecedent influencing factors related to firms‟ internal and external environments. Applying this framework, hypotheses are developed and tested using multivariate models and a large sample of hand-collected MEFs (7,213) issued by top 500 ASX-listed companies over the 1994 to 2008 period. The results reveal strong support for the hypotheses. First, MEFs are issued to reduce information asymmetry, litigation risk and signal superior performance. Second, firms with better financial performance, smaller earnings changes, and lower operating uncertainty provide better quality MEFs. Third, forecast frequency and quality (accuracy, timeliness and precision) noticeably improve as litigation risk increases. However, managers appear to be still reluctant to disclose earnings forecasts when there are large earnings changes, and an asymmetric treatment of news type continues to prevail (a good news bias). Thus, the findings generally provide support for the effectiveness of the CDR regulatory reforms in improving disclosure behaviour and will be valuable to market participants and corporate regulators in understanding the implications of management forecasting decisions and areas for further improvement.

Building a USDL repository as a service using MDE

The decision to represent the USDL abstract syntax as a metamodel, shown as a set of UML diagrams, has two main benefits: the ability to show a well- understood standard graphical representation of the concepts and their relation- ships to one another, and the ability to use object-oriented frameworks such as Eclipse Modeling Framework (EMF) to assist in the automated generation of tool support for USDL service descriptions.

Constructing tag ontology from folksonomy based on WordNet

With the emergence of Web 2.0, Web users can classify Web items of their interest by using tags. Tags reflect users’ understanding to the items collected in each tag. Exploring user tagging behavior provides a promising way to understand users’ information needs. However, free and relatively uncontrolled vocabulary has its drawback in terms of lack of standardization and semantic ambiguity. Moreover, the relationships among tags have not been explored even there exist rich relationships among tags which could provide valuable information for us to better understand users. In this paper, we propose a novel approach to construct tag ontology based on the widely used general ontology WordNet to capture the semantics and the structural relationships of tags. Ambiguity of tags is a challenging problem to deal with in order to construct high quality tag ontology. We propose strategies to find the semantic meanings of tags and a strategy to disambiguate the semantics of tags based on the opinion of WordNet lexicographers. In order to evaluate the usefulness of the constructed tag ontology, in this paper we apply the extracted tag ontology in a tag recommendation experiment. We believe this is the first application of tag ontology for recommendation making. The initial result shows that by using the tag ontology to re-rank the recommended tags, the accuracy of the tag recommendation can be improved.

Evaluation framework for focused link discovery

Information has no value unless it is accessible. Information must be connected together so a knowledge network can then be built. Such a knowledge base is a key resource for Internet users to interlink information from documents. Information retrieval, a key technology for knowledge management, guarantees access to large corpora of unstructured text. Collaborative knowledge management systems such as Wikipedia are becoming more popular than ever; however, their link creation function is not optimized for discovering possible links in the collection and the quality of automatically generated links has never been quantified. This research begins with an evaluation forum which is intended to cope with the experiments of focused link discovery in a collaborative way as well as with the investigation of the link discovery application. The research focus was on the evaluation strategy: the evaluation framework proposal, including rules, formats, pooling, validation, assessment and evaluation has proved to be efficient, reusable for further extension and efficient for conducting evaluation. The collection-split approach is used to re-construct the Wikipedia collection into a split collection comprising single passage files. This split collection is proved to be feasible for improving relevant passages discovery and is devoted to being a corpus for focused link discovery. Following these experiments, a mobile client-side prototype built on iPhone is developed to resolve the mobile Search issue by using focused link discovery technology. According to the interview survey, the proposed mobile interactive UI does improve the experience of mobile information seeking. Based on this evaluation framework, a novel cross-language link discovery proposal using multiple text collections is developed. A dynamic evaluation approach is proposed to enhance both the collaborative effort and the interacting experience between submission and evaluation. A realistic evaluation scheme has been implemented at NTCIR for cross-language link discovery tasks.

Tensor query expansion : a cognitively motivated relevance model

In information retrieval, a user's query is often not a complete representation of their real information need. The user's information need is a cognitive construction, however the use of cognitive models to perform query expansion have had little study. In this paper, we present a cognitively motivated query expansion technique that uses semantic features for use in ad hoc retrieval. This model is evaluated against a state-of-the-art query expansion technique. The results show our approach provides significant improvements in retrieval effectiveness for the TREC data sets tested.

Trading spaces : on the lore and limitations of latent semantic analysis

Two decades after its inception, Latent Semantic Analysis(LSA) has become part and parcel of every modern introduction to Information Retrieval. For any tool that matures so quickly, it is important to check its lore and limitations, or else stagnation will set in. We focus here on the three main aspects of LSA that are well accepted, and the gist of which can be summarized as follows: (1) that LSA recovers latent semantic factors underlying the document space, (2) that such can be accomplished through lossy compression of the document space by eliminating lexical noise, and (3) that the latter can best be achieved by Singular Value Decomposition. For each aspect we performed experiments analogous to those reported in the LSA literature and compared the evidence brought to bear in each case. On the negative side, we show that the above claims about LSA are much more limited than commonly believed. Even a simple example may show that LSA does not recover the optimal semantic factors as intended in the pedagogical example used in many LSA publications. Additionally, and remarkably deviating from LSA lore, LSA does not scale up well: the larger the document space, the more unlikely that LSA recovers an optimal set of semantic factors. On the positive side, we describe new algorithms to replace LSA (and more recent alternatives as pLSA, LDA, and kernel methods) by trading its l2 space for an l1 space, thereby guaranteeing an optimal set of semantic factors. These algorithms seem to salvage the spirit of LSA as we think it was initially conceived.

An effective approach to verbose queries using a limited dependencies language model

Intuitively, any ‘bag of words’ approach in IR should benefit from taking term dependencies into account. Unfortunately, for years the results of exploiting such dependencies have been mixed or inconclusive. To improve the situation, this paper shows how the natural language properties of the target documents can be used to transform and enrich the term dependencies to more useful statistics. This is done in three steps. The term co-occurrence statistics of queries and documents are each represented by a Markov chain. The paper proves that such a chain is ergodic, and therefore its asymptotic behavior is unique, stationary, and independent of the initial state. Next, the stationary distribution is taken to model queries and documents, rather than their initial distributions. Finally, ranking is achieved following the customary language modeling paradigm. The main contribution of this paper is to argue why the asymptotic behavior of the document model is a better representation then just the document’s initial distribution. A secondary contribution is to investigate the practical application of this representation in case the queries become increasingly verbose. In the experiments (based on Lemur’s search engine substrate) the default query model was replaced by the stable distribution of the query. Just modeling the query this way already resulted in significant improvements over a standard language model baseline. The results were on a par or better than more sophisticated algorithms that use fine-tuned parameters or extensive training. Moreover, the more verbose the query, the more effective the approach seems to become.