Class Of Query Research Articles

Content-based image retrieval based on supervised learning and statistical-based moments

Content-based image retrieval (CBIR) system generally retrieves images based on the matching of the query image from all the images of the database. This exhaustive matching and searching slow down the image retrieval process. In this paper, a fast and effective CBIR system is proposed which uses supervised learning-based image management and retrieval techniques. It utilizes machine learning approaches as a prior step for speeding up image retrieval in the large database. For the implementation of this, first, we extract statistical moments and the orthogonal-combination of local binary patterns (OC-LBP)-based computationally light weighted color and texture features. Further, using some ground truth annotation of images, we have trained the multi-class support vector machine (SVM) classifier. This classifier works as a manager and categorizes the remaining images into different libraries. However, at the query time, the same features are extracted and fed to the SVM classifier. SVM detects the class of query and searching is narrowed down to the corresponding library. This supervised model with weighted Euclidean Distance (ED) filters out maximum irrelevant images and speeds up the searching time. This work is evaluated and compared with the conventional model of the CBIR system on two benchmark databases, and it is found that the proposed work is significantly encouraging in terms of retrieval accuracy and response time for the same set of used features.

Supporting the discoverability of open educational resources

Open Educational Resources (OERs), now available in large numbers, have a considerable potential to improve many aspects of society, yet one of the factors limiting this positive impact is the difficulty to discover them. This study investigates and proposes strategies to better support educators in discovering OERs, mainly focusing on secondary education. The literature suggests that the effectiveness of existing search systems could be improved by supporting high-level and domain-oriented tasks. Hence a preliminary taxonomy of discovery-related tasks was developed, based on the analysis of the literature, interpreted through Information Foraging Theory. This taxonomy was empirically evaluated with a few experienced educators, to preliminary identify an interesting class of Query By Examples (QBE) expansion by similarity tasks, which avoids the need to decompose natural high-level tasks in a complex sequence of sub-tasks. Following the Design Science Research methodology, three prototypes to support as well as to refine those tasks were iteratively designed, implemented, and evaluated involving an increasing number of educators in usability oriented studies. The resulting high-level and domain-oriented blended search/recommendation strategy, transparently replicates Google searches in specialized networks, and identifies similar resources with a QBE strategy. It makes use of a domain-oriented similarity metric based on shared schema.org/LRMI alignments to educational frameworks, and clusters results in expandable classes of comparable degree of similarity. The summative evaluation shows that educators appreciate this exploratory-oriented strategy because – balancing similarity and diversity – it supports their high-level tasks, such as lesson planning and personalization of education.

An access and inference control model for time series databases

Today, many applications produce and use time series data. The data of this type may contain sensitive information. So they should be protected against unauthorized accesses. In this paper, security issues of time series data are identified and an access and inference control model for satisfying the identified security requirements is proposed. Using this model, administrators can define authorization rules based on various time-based granularities (e.g. day or month) and apply value-based constraints over the accessed times series data. Furthermore, they can define policy rules over the composition of multiple time-series other than the base time-series data. Detecting and resolving different types of conflicts between the simple, aggregation, or composition access rules over the time series data is a challenging issue which is investigated in this paper. Detecting explicit and implicit conflicts in this model prevents information inference from hierarchical time series data. To prove the applicability of the model, a reference monitor based on the proposed model has been implemented as a secure access layer on top of OpenTSDB (a time series database). Our evaluation shows that the overhead imposed on queries by the secure access layer depends on several parameters such as class of query, number of authorization rules, complexity of value constraints, and number of data being accessed. In our experiments, the overhead was variable in the range of 4% in the best case to 184% in the worst case.

On-the-fly learning for visual search of large-scale image and video datasets.

The objective of this work is to visually search large-scale video datasets for semantic entities specified by a text query. The paradigm we explore is constructing visual models for such semantic entities on-the-fly, i.e. at run time, by using an image search engine to source visual training data for the text query. The approach combines fast and accurate learning and retrieval, and enables videos to be returned within seconds of specifying a query. We describe three classes of queries, each with its associated visual search method: object instances (using a bag of visual words approach for matching); object categories (using a discriminative classifier for ranking key frames); and faces (using a discriminative classifier for ranking face tracks). We discuss the features suitable for each class of query, for example Fisher vectors or features derived from convolutional neural networks (CNNs), and how these choices impact on the trade-off between three important performance measures for a real-time system of this kind, namely: (1) accuracy, (2) memory footprint, and (3) speed. We also discuss and compare a number of important implementation issues, such as how to remove ‘outliers’ in the downloaded images efficiently, and how to best obtain a single descriptor for a face track. We also sketch the architecture of the real-time on-the-fly system. Quantitative results are given on a number of large-scale image and video benchmarks (e.g. TRECVID INS, MIRFLICKR-1M), and we further demonstrate the performance and real-world applicability of our methods over a dataset sourced from 10,000 h of unedited footage from BBC News, comprising 5M+ key frames.

AN OPTIMIZED FEATURE EXTRACTION TECHNIQUE FOR CONTENT BASED IMAGE RETRIEVAL

Content-based image retrieval (CBIR) is an active research area with the development of multimedia technologies and has become a source of exact and fast retrieval. The aim of CBIR is to search and retrieve images from a large database and find out the best match for the given query. Accuracy and efficiency for high dimensional datasets with enormous number of samples is a challenging arena. In this paper, Content Based Image Retrieval using various features such as color, shape, texture is made and a comparison is made among them. The performance of the retrieval system is evaluated depending upon the features extracted from an image. The performance was evaluated using precision and recall rates. Haralick texture features were analyzed at 0 o, 45 o, 90 o, 180 o using gray level co-occurrence matrix. Color feature extraction was done using color moments. Structured features and multiple feature fusion are two main technologies to ensure the retrieval accuracy in the system. GIST is considered as one of the main structured features. It was experimentally observed that combination of these techniques yielded superior performance than individual features. The results for the most efficient combination of techniques have also been presented and optimized for each class of query.

A Density based Priority Queue Strategy to Evaluate Iceberg Queries Efficiently using Compressed Bitmap Indices

iceberg query is a special class of aggregation query that computes aggregated values upon user interested threshold (T).The bitmap index is a common data structure for fast retrieval of matching tuples from data base table. These resultant tuples are useful to compute aggregations such as SUM, COUNT, AVG, MIN, MAX, and RANK. In this paper, we propose a density based bitmap pruning strategy to evaluate iceberg queries efficiently using compressed bitmap indices. The strategy prioritizes the vectors to be enter in to priority queue by allowing high density of 1's count that achieve optimal pruning effect. Extensive experimentation demonstrates our proposed approach is much more efficient than existing strategy.

An Algorithm to Evaluate Iceberg Query using Compacted Bitmap Vector

The data storing and retrieving are playing a major role in the data clustering and data warehousing techniques. The effectiveness of a data retrieving method depends upon the data specific queries for retrieving the data from the database. Iceberg query is a unique class of aggregation query, which computes aggregate values above a given threshold. Many data mining queries are basically ice berg queries. The major part taken into the consideration about the AND operation in the iceberg queries. The reduced number of AND operation increases the effectiveness of the iceberg query. In this work, an efficient iceberg query evaluation process is proposed by reducing the bitwise AND operations needed to find the item pairs. In the proposed approach two solutions are introduced to reduce the bitwise AND operations. Randomly identifying 'N' 1-bit positions instead of first 1-bit position and reducing the zero bit values from the Most Significant Side so that the bit map vector will be reduced in such a way, the bitwise operations needed is reduced. The experimentation is conducted on two datasets in order to evaluate the performance of the proposed iceberg query evaluation algorithm. In the case of retail datasets, the time required for processing 100000 tuples is 864 milliseconds. On the other hand, the same for synthetic dataset, T10I4D100K, is 910 milliseconds.

Comparative Study and Optimization of Feature-Extraction Techniques for Content based Image Retrieval

The aim of a Content-Based Image Retrieval (CBIR) system, also known as Query by Image Content (QBIC), is to help users to retrieve relevant images based on their contents. CBIR technologies provide a method to find images in large databases by using unique descriptors from a trained image. The image descriptors include texture, color, intensity and shape of the object inside an image. Several feature-extraction techniques viz., Average RGB, Color Moments, Co-occurrence, Local Color Histogram, Global Color Histogram and Geometric Moment have been critically compared in this paper. However, individually these techniques result in poor performance. So, combinations of these techniques have also been evaluated and results for the most efficient combination of techniques have been presented and optimized for each class of image query. We also propose an improvement in image retrieval performance by introducing the idea of Query modification through image cropping. It enables the user to identify a region of interest and modify the initial query to refine and personalize the image retrieval results.

A hierarchical approach to protein fold prediction.

Fold recognition, assigning novel proteins to known structures, forms an important component of the overall protein structure discovery process. The available methods for protein fold recognition are limited by the low fold-coverage and/or low prediction accuracies. We describe here a new Support Vector Machine (SVM)-based method for protein fold prediction with high prediction accuracy and high fold-coverage. The new method of fold prediction with high fold-coverage was developed by training and testing on a large number of folds in order to make the method suitable for large scale fold predictions. However, presence of large number of folds in the training set made the classification task difficult as a consequence of increased complexity involved in binary classifications of SVMs. In order to overcome this complexity we adopted a hierarchical approach where fold-prediction is made in two steps. At the first step structural class of the query is predicted and at the second step fold is predicted within the predicted structural class. This decreased the complexity of the classification problem and also improved the overall fold prediction accuracy. To the best of our knowledge this is the first taxonomic fold recognition method to cover over 700 protein-folds and gives prediction accuracy of around 70% on a benchmark dataset. Since the new method gives rise to state of the art prediction performance and hence can be very useful for structural characterization of proteins discovered in various genomes.

웹 정보검색 시스템의 문서 순위 결정

인터넷의 발달로 인해 웹에서 얻을 수 있는 정보의 종류와 수는 급진적으로 증가하고 있다. 기존의 문서 위주의 구성에서 멀티미디어 서비스, 쇼핑몰 등 종류와 매체에 있어서 다양한 변화를 보이고 있다. 이에 따라 사용자가 요구하는 정보의 단위는 문서 뿐만 아니라, 사이트 그리고 서비스 단위로 확장되고 있다. 웹 환경에서 사용자의 정보 요구를 보면 크게 세가지로 볼 수 있다. 첫째는 원하는 정보를 설명하는 혹은 정보와 관련된 문서를 찾는 내용검색, 둘째는 사용자가 관심 있어 하는 개인이나 단체의 사이트 입구를 찾는 사이트 검색, 셋째는 사용자가 관심 있어 하는 서비스를 제공하는 웹 페이지를 찾는 서비스 검색을 들 수 있다. 본 논문에서는 이러한 사용자의 정보 요구 목적에 따라서 문서 순위화가 달라져야 함을 보인다. 지금까지 정보 검색에서 언급된 내용 정보, 링크 정보 그리고 URL 정보의 유용함을 사용자의 정보 요구 형태에 따라서 분류한다. 내용 검색에서는 내용 정보가 유용한 반면 링크 정보와 URL 정보를 결합할 경우 성능의 저하를 초래했다. 반면 사이트 검색에서는 내용 정보만 쓰는 것 보다는 링크 정보와 URL 정보를 결합할 경우 성능의 향상을 얻을 수 있었다. The Web is rich with various sources of information. It contains the contents of documents, multimedia data, shopping materials and so on. Due to the massive and heterogeneous web document collections, users want to find various types of target pages. We can classify user queries as three categories according to users'intent, content search, the site search, and the service search. In this paper, we present that different strategies are needed to meet the need of a user. Also we show the properties of content information, link information and URL information according to the class of a user query. In the content search, content information showed the good result. However, we lost the performance by combining link information and URL information. In the site search, we could increase the performance by combining link information and URL information.

On supporting containment queries in relational database management systems

Virtually all proposals for querying XML include a class of query we term “containment queries”. It is also clear that in the foreseeable future, a substantial amount of XML data will be stored in relational database systems. This raises the question of how to support these containment queries. The inverted list technology that underlies much of Information Retrieval is well-suited to these queries, but should we implement this technology (a) in a separate loosely-coupled IR engine, or (b) using the native tables and query execution machinery of the RDBMS? With option (b), more than twenty years of work on RDBMS query optimization, query execution, scalability, and concurrency control and recovery immediately extend to the queries and structures that implement these new operations. But all this will be irrelevant if the performance of option (b) lags that of (a) by too much. In this paper, we explore some performance implications of both options using native implementations in two commercial relational database systems and in a special purpose inverted list engine. Our performance study shows that while RDBMSs are generally poorly suited for such queries, under conditions they can outperform an inverted list engine. Our analysis further identifies two significant causes that differentiate the performance of the IR and RDBMS implementations: the join algorithms employed and the hardware cache utilization. Our results suggest that contrary to most expectations, with some modifications, a native implementations in an RDBMS can support this class of query much more efficiently.

Recursive query plans for data integration

Generating query-answering plans for data integration systems requires to translate a user query, formulated in terms of a mediated schema, to a query that uses relations that are actually stored in data sources. Previous solutions to the translation problem produced sets of conjunctive plans, and were therefore limited in their ability to handle recursive queries and to exploit data sources with binding-pattern limitations and functional dependencies that are known to hold in the mediated schema. As a result, these plans were incomplete w.r.t. sources encountered in practice (i.e., produced only a subset of the possible answers). We describe the novel class of recursive query answering plans, which enables us to settle three open problems. First, we describe an algorithm for finding a query plan that produces the maximal set of answers from the sources for arbitrary recursive queries. Second, we extend this algorithm to use the presence of functional and full dependencies in the mediated schema. Third, we describe an algorithm for finding the maximal query plan in the presence of binding-pattern restrictions in the sources. In all three cases, recursive plans are necessary in order to obtain a maximal query plan.

CATEGORISATION OF COMPUTABLE DATABASE QUERIES

We present an alternative approach to that of Chandra and Harel [5] and Abiteboul and Vianu [1] in considering computable database queries, which are mappings from sets of records to sets of records. In particular, we view a computable database query as being realised via a Turing-computable mapping from strings to strings and an encoding, which encodes the input set of records into an appropriate string. An encoding of a set of records consists of two components: an ordering function, which orders the records in a set as well as the values of each record in the set, and an isomorphism, which maps the values in the records of the set to strings. An important class of encodings, called free encodings, whose isomorphism has the same semantics as the identity mapping on record values, is also defined. Our analysis of computable database queries elucidates the notion of a computable database query by dealing with the problem of how a database language can be implemented on a standard Turing machine that does not cater directly for mappings from sets of records to sets of records. We carry out our analysis by categorising computable database queries into subclasses and by establishing the relationships that exist amongst these subclasses. We also investigate an equivalence relation on computable database queries; two computable database queries are related if they are realised via the same Turing-computable mapping, say d. We prove the following interesting result regarding the cardinality of the equivalence class of a computable query with respect to the said equivalence relation: either d does not realise any computable query, or d realises exactly one computable query, or d realises a countably infinite set of computable queries. Our final result shows that, by adding membership queries to the class of encoding-independent computable queries, the closure of the resulting extended class under composition of mappings is the set of all isomorphism-independent computable queries.

An Overview of Information Retrieval Subjects

The aim of an information retrieval system is to find information items relevant to an information need. As relevance is a kind of similarity relation between the concepts represented by the information item and those represented by the formulation of the information need, it is not astonishing to discover that the class of possible query forms -formulations of the information needs - is the same as the class of possible representations of information items. This article overviews current research problems in information structure and query evaluation