All rights reserved. During times of crisis microblogging platforms such as Twitter have played an important role as a communication channel to distribute information. Particularly, disaster-related tweets are valuable resources when tagged with their location for detecting unexpected events. However, they often contain different types of location and one of the main challenges is resolving the ambiguity involved in their locations. The process of identifying phrase portions in unstructured texts with possible spatial aspects and disambiguating these references by linking them to geographic coordinates is known as Geotagging. In the context of crisis management, this paper presents OzCT geotagger that automatically detects the location(s) mentioned in the content of tweets with three possibilities: definite, ambiguous and no-location. It also semantically annotates the tweet components utilizing existing and new ontologies. The OzCT geotagger has been recently deployed in a trial system of the OzCrisisTracker application. Experiments demonstrate that the precision and recall for detection of the definite locations against geotagging by human judgement are on average of 80%. We also conclude that the accuracy of geographical focus of the OzCT geotagger is considerably higher than other systems. While existing geocoding systems have lower coverage for suburb and street focus, our approach detects suburbs in more than 60% situations.

The success of an enterprise information retrieval system is determined by interactions among three key entities: the search engine employed; the service provider who delivers, modifies, and maintains the engine; and the users of the service within the organization. Evaluations of an enterprise search have predominately focused on the effectiveness and efficiency of the engine, with very little analysis of user involvement in the process, and none on the role of service providers. We propose and evaluate a model of costs and benefits to a service provider when investing in enhancements to the ranking of documents returned by their search engine. We demonstrate the model through a case study to analyze the potential impact of using domain experts to provide enhanced mediated search results. By demonstrating how to quantify the cost and benefit of an improved information retrieval system to the service provider, our case study shows that using the relevance assessments of domain experts to rerank original search results can significantly improve the accuracy of ranked lists. Moreover, the service provider gains substantial return on investment and a higher search success rate by investing in the relevance assessments of domain experts. Our cost and benefit analysis results are contrasted with standard modes of effectiveness analysis, including quantitative (using measures such as precision) and qualitative (through user preference surveys) approaches. Modeling costs and benefits explicitly can provide useful insights that the other approaches do not convey.

Semantic models help in achieving semantic interoperability among sources of data and applications. The necessity to efficiently manage these types of objects has increased the number of specialized repositories, usually referred to as semantic databases. An increasing number of project initiatives have been recorded that choose to formalize application knowledge using ontologies and semantic data representation. Due to the various sensitivities of data, suitable access control mechanisms pertaining to the semantic repository should be put in place to ensure that only authorized users can obtain access to the information in its entirety. In fact, deciding what can be made available to the user without revealing confidential information is made even more difficult because the user may be able to apply logic and reasoning to infer confidential information from the knowledge being provided. In this paper, we design an authorization security model enforced on a semantic model's entities (concepts) and also propagate on their individuals in the OWL database through an inference policy engine. We provide TBox access control for the construction of a TBox family and propagate this based on the construction of concept taxonomies. We also provide ABox label-based access control for facts in the domain knowledge and report experiments to evaluate the effects of access control on reasoning and modularization.

Maintaining strictness in dimensions is important in integration of data warehouses. A dimension that satisfies all of its roll-up constraints is said to be strict, a property that is required for correct aggregation. Existing work on instance matching does not address the problem of enforcing the strictness of roll-up constraints. In this paper, we use a graph matching-based approach to dimension instance matching and propose an algorithm that enforces strictness and reduces false positives. Making use of similarity flooding, the graph matching algorithm can be greedy in identifying matching members, we propose heuristics to further reduce false positive matches and reduce false strictness. Experiments on real-world data demonstrates the effectiveness of our proposed approach.

Service matching approaches trade precision for recall, creating the need for users to choose the correct services, which obviously is a major obstacle for automating the service discovery and aggregation processes. Our approach to overcome this problem, is to eliminate the appearance of false positives by returning only the correct services. As different users have different semantics for what is correct, we argue that the correctness of the matching results must be determined according to the achievement of users' goals: that only services achieving users' goals are considered correct. To determine such correctness, we argue that the matching process should be based primarily on the high-level functional specifications (namely goals, achievement contexts, and external behaviors). In this article, we propose models, data structures, algorithms, and theorems required to correctly match such specifications. We propose a model calledG+, to capture such specifications, for both services and users, in a machine-understandable format. We propose a data structure, called a Concepts Substitutability Graph (CSG), to capture the substitution semantics of application domain concepts in a context-based manner, in order to determine the semantic-preserving mapping transformations required to match differentG+models. We also propose a behavior matching approach that is able to match states in an m-to-n manner, such that behavior models with different numbers of state transitions can be matched. Finally, we show how services are matched and aggregated according to theirG+models. Results of supporting experiments demonstrate the advantages of the proposed service matching approaches.

Ontology evaluation is a maturing discipline with methodologies and measures being developed and proposed. However, evaluation methods that have been proposed have not been applied to specific examples. In this paper, we present the state-of-the-art in ontology evaluation - current methodologies, criteria and measures, analyse appropriate evaluations that are important to our application - browsing in Wikipedia, and apply these evaluations in the context of ontologies with varied properties. Specifically, we seek to evaluate ontologies based on categories found in Wikipedia.

The application of machine learning techniques to image and video search has been shown to boost the performance of multimedia retrieval systems, and promises to lead to more generalized semantic search approaches. In particular, the availability of large training collections allows model-driven search using a substantial number of semantic concepts. The training collections are obtained in a manual annotation process where human raters review images and assign predefined semantic concept labels. Besides being prone to human error, manual image annotation is biased by the view of the individual annotator because visual information almost always leaves room for ambiguity. Ideally, several independent judgments are obtained per image, and the inter-rater agreement is assessed. While disagreement between ratings bears valuable information on the annotation quality, it complicates the task of clearly classifying rated images based on multiple judgments. In the absence of a gold standard, evaluating multiple judgments and resolving disagreement between raters is not trivial. In this paper, we present an approach using latent structure analysis to solve this problem. We apply latent class modeling to the annotation data collected during the TRECVID 2005 Annotation Forum, and demonstrate how to use this statistic to clearly classify each image on the basis of varying numbers of ratings. We use latent class modeling to quantify the annotation quality and discuss the results in comparison with the well-known Kappa inter-rater agreement measure.

Content based retrieval systems have been developed for querying image data in which the users can pose queries based on visual properties such as color and texture. These systems, which have advanced the state of the art in image database systems, remarkably lack formal query languages. The traditional query languages are unable to capture the inherent fuzzy nature of the image data and content based querying. The fuzzy object query language (FOQL) presented in this paper addresses the need to support fuzzy values and fuzzy collections required for image databases. It can be used for defining schemas and high level concepts, and for querying image databases. It captures the inherent fuzzy nature of content based retrieval by keeping the query results fuzzy as against other query languages. The users can interactively refine their queries and high level concept definitions using recursive and named query definition constructs in FOQL. Being an extension of ODMG-OQL, FOQL can be easily mapped to ODMG-compliant visual query languages.

No abstract available.

Advanced database applications require facilities such as text indexing, image storage, and the ability to store data with a complex structure. However, these facilities are not usually included in traditional database systems. In this paper we describe Atlas, a nested relational database system that has been designed for text-based applications. The Atlas query language is TQL, an SQL-like query language with text operators. The query language is supported by signature file text indexing techniques, and by a parser that can be configured for different text formats and even some foreign languages. Atlas can also be used to store images and audio.>

It is common to model the distribution of words in text by measures such as the Poisson approximation. However, these measures ignore effects such as clustering: our analysis of document collections demonstrates that the Poisson approximation can significantly overestimate the probability that a document contains a word. Based on our analysis, we propose a new model for distribution of words in text, and show how this model can be used to estimate the probability that a document contains a word and the number of distinct words in a document. © 1992 John Wiley & Sons, Inc.