Um Mecanismo de realimentação de relevâncias para recuperação de informações visuais utilizando feições a partir de imagens JPEG

Esta dissertação propõe e discute um mecanismo de realimentação de relevâncias (i. e. “Relevance Feedback”). A técnica de realimentação de relevâncias foi introduzida inicialmente em meados dos anos 60, como uma estratégia para refinamento de consultas para a recuperação de informações. Como uma técnica de refinamento de consultas, foi aplicada inicialmente em sistemas de recuperação de informações textuais. Neste caso, os termos ou expressões consideradas importantes, são utilizados na formulação de uma nova consulta. Ao surgirem os sistemas de recuperação de informação visual baseada em conteúdo (CBVIR), houve a necessidade de serem introduzidos novos elementos associados a esse processo de reformulação de consultas, de tal forma que fossem utilizados não apenas as informações de alto nível, como os termos e expressões. Esses novos elementos passaram a considerar também a subjetividade de percepção humana em relação ao conteúdo visual. Neste trabalho, apresenta-se um processo de extração e representação desse conteúdo, através da utilização de feições (conteúdo) de cor e textura, extraídos de imagens JPEG, uma vez que no processo de compressão de imagens nesse formato, utiliza-se coeficientes da Transformada Discreta do Cosseno (DCT), sendo, portanto esses coeficientes utilizados como elementos que possuem as informações associadas a cor e textura na imagem. Além da DCTé utilizada a Transformação Mandala [YSH 83] no processo de agrupamento de somente 10 coeficientes, com o objetivo de produzir 10 imagens com resoluça menor que a imagem original, mas que representam cada uma, o conteúdo de uma frequência particular da imagem original. A escolha por uma representação como essa,é a garantia de uma redução significativa na quantidade de dados a serem processados. Entretanto, a representação obtida nesse formato para as imagens,é com base em conteúdo global de cor e textura, o que pode produzir resultados insatisfatórios. A introdução de um mecanismo de realimentação de relevâncias, associado à representação utilizada, permite contornar a dificuldade apontada acima, através da obtenção de consultas subsequentes, selecionando os objetos mais relevantes, assim como menos objetos não relevantes, utilizando o conhecimento do usuário de forma interativa no refinamento de consultas para recuperação de informações visuais.

Recuperação de informação de música e dados ID3: possíveis aplicações

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

PRoSPer: perceptual similarity queries in medical CBIR systems through user profiles

In this paper, we present a novel approach to perform similarity queries over medical images, maintaining the semantics of a given query posted by the user. Content-based image retrieval systems relying on relevance feedback techniques usually request the users to label relevant/irrelevant images. Thus, we present a highly effective strategy to survey user profiles, taking advantage of such labeling to implicitly gather the user perceptual similarity. The profiles maintain the settings desired for each user, allowing tuning of the similarity assessment, which encompasses the dynamic change of the distance function employed through an interactive process. Experiments on medical images show that the method is effective and can improve the decision making process during analysis.

Query rewriting in information retrieval: automatic context extraction from local user documents to improve query results

The central objective of research in Information Retrieval (IR) is to discover new techniques to retrieve relevant information in order to satisfy an Information Need. The Information Need is satisfied when relevant information can be provided to the user. In IR, relevance is a fundamental concept which has changed over time, from popular to personal, i.e., what was considered relevant before was information for the whole population, but what is considered relevant now is specific information for each user. Hence, there is a need to connect the behavior of the system to the condition of a particular person and his social context; thereby an interdisciplinary sector called Human-Centered Computing was born. For the modern search engine, the information extracted for the individual user is crucial. According to the Personalized Search (PS), two different techniques are necessary to personalize a search: contextualization (interconnected conditions that occur in an activity), and individualization (characteristics that distinguish an individual). This movement of focus to the individual's need undermines the rigid linearity of the classical model overtaken the ``berry picking'' model which explains that the terms change thanks to the informational feedback received from the search activity introducing the concept of evolution of search terms. The development of Information Foraging theory, which observed the correlations between animal foraging and human information foraging, also contributed to this transformation through attempts to optimize the cost-benefit ratio. This thesis arose from the need to satisfy human individuality when searching for information, and it develops a synergistic collaboration between the frontiers of technological innovation and the recent advances in IR. The search method developed exploits what is relevant for the user by changing radically the way in which an Information Need is expressed, because now it is expressed through the generation of the query and its own context. As a matter of fact the method was born under the pretense to improve the quality of search by rewriting the query based on the contexts automatically generated from a local knowledge base. Furthermore, the idea of optimizing each IR system has led to develop it as a middleware of interaction between the user and the IR system. Thereby the system has just two possible actions: rewriting the query, and reordering the result. Equivalent actions to the approach was described from the PS that generally exploits information derived from analysis of user behavior, while the proposed approach exploits knowledge provided by the user. The thesis went further to generate a novel method for an assessment procedure, according to the "Cranfield paradigm", in order to evaluate this type of IR systems. The results achieved are interesting considering both the effectiveness achieved and the innovative approach undertaken together with the several applications inspired using a local knowledge base.

MIRACLE at Ad-Hoc CLEF 2005: Merging and Combining Without Using a Single Approach

This paper presents the 2005 Miracle’s team approach to the Ad-Hoc Information Retrieval tasks. The goal for the experiments this year was twofold: to continue testing the effect of combination approaches on information retrieval tasks, and improving our basic processing and indexing tools, adapting them to new languages with strange encoding schemes. The starting point was a set of basic components: stemming, transforming, filtering, proper nouns extraction, paragraph extraction, and pseudo-relevance feedback. Some of these basic components were used in different combinations and order of application for document indexing and for query processing. Second-order combinations were also tested, by averaging or selective combination of the documents retrieved by different approaches for a particular query. In the multilingual track, we concentrated our work on the merging process of the results of monolingual runs to get the overall multilingual result, relying on available translations. In both cross-lingual tracks, we have used available translation resources, and in some cases we have used a combination approach.

MIRACLE’s Hybrid Approach to Bilingual and Monolingual Information Retrieval

The main goal of the bilingual and monolingual participation of the MIRACLE team in CLEF 2004 was to test the effect of combination approaches on information retrieval. The starting point was a set of basic components: stemming, transformation, filtering, generation of n-grams, weighting and relevance feedback. Some of these basic components were used in different combinations and order of application for document indexing and for query processing. A second order combination was also tested, mainly by averaging or selective combination of the documents retrieved by different approaches for a particular query.

A scalable approach for content based image retrieval in cloud datacenter

The emergence of cloud datacenters enhances the capability of online data storage. Since massive data is stored in datacenters, it is necessary to effectively locate and access interest data in such a distributed system. However, traditional search techniques only allow users to search images over exact-match keywords through a centralized index. These techniques cannot satisfy the requirements of content based image retrieval (CBIR). In this paper, we propose a scalable image retrieval framework which can efficiently support content similarity search and semantic search in the distributed environment. Its key idea is to integrate image feature vectors into distributed hash tables (DHTs) by exploiting the property of locality sensitive hashing (LSH). Thus, images with similar content are most likely gathered into the same node without the knowledge of any global information. For searching semantically close images, the relevance feedback is adopted in our system to overcome the gap between low-level features and high-level features. We show that our approach yields high recall rate with good load balance and only requires a few number of hops.

Recuperación cruzada de información pública clínico-genómica a partir de consultas sobre repositorios clínicos privados

Parte de la investigación biomédica actual se encuentra centrada en el análisis de datos heterogéneos. Estos datos pueden tener distinto origen, estructura, y semántica. Gran cantidad de datos de interés para los investigadores se encuentran en bases de datos públicas, que recogen información de distintas fuentes y la ponen a disposición de la comunidad de forma gratuita. Para homogeneizar estas fuentes de datos públicas con otras de origen privado, existen diversas herramientas y técnicas que permiten automatizar los procesos de homogeneización de datos heterogéneos. El Grupo de Informática Biomédica (GIB) [1] de la Universidad Politécnica de Madrid colabora en el proyecto europeo P-medicine [2], cuya finalidad reside en el desarrollo de una infraestructura que facilite la evolución de los procedimientos médicos actuales hacia la medicina personalizada. Una de las tareas enmarcadas en el proyecto P-medicine que tiene asignado el grupo consiste en elaborar herramientas que ayuden a usuarios en el proceso de integración de datos contenidos en fuentes de información heterogéneas. Algunas de estas fuentes de información son bases de datos públicas de ámbito biomédico contenidas en la plataforma NCBI [3] (National Center for Biotechnology Information). Una de las herramientas que el grupo desarrolla para integrar fuentes de datos es Ontology Annotator. En una de sus fases, la labor del usuario consiste en recuperar información de una base de datos pública y seleccionar de forma manual los resultados relevantes. Para automatizar el proceso de búsqueda y selección de resultados relevantes, por un lado existe un gran interés en conseguir generar consultas que guíen hacia resultados lo más precisos y exactos como sea posible, por otro lado, existe un gran interés en extraer información relevante de elevadas cantidades de documentos, lo cual requiere de sistemas que analicen y ponderen los datos que caracterizan a los mismos. En el campo informático de la inteligencia artificial, dentro de la rama de la recuperación de la información, existen diversos estudios acerca de la expansión de consultas a partir de retroalimentación relevante que podrían ser de gran utilidad para dar solución a la cuestión. Estos estudios se centran en técnicas para reformular o expandir la consulta inicial utilizando como realimentación los resultados que en una primera instancia fueron relevantes para el usuario, de forma que el nuevo conjunto de resultados tenga mayor proximidad con los que el usuario realmente desea. El objetivo de este trabajo de fin de grado consiste en el estudio, implementación y experimentación de métodos que automaticen el proceso de extracción de información trascendente de documentos, utilizándola para expandir o reformular consultas. De esta forma se pretende mejorar la precisión y el ranking de los resultados asociados. Dichos métodos serán integrados en la herramienta Ontology Annotator y enfocados a la fuente de datos de PubMed [4].---ABSTRACT---Part of the current biomedical research is focused on the analysis of heterogeneous data. These data may have different origin, structure and semantics. A big quantity of interesting data is contained in public databases which gather information from different sources and make it open and free to be used by the community. In order to homogenize thise sources of public data with others which origin is private, there are some tools and techniques that allow automating the processes of integration heterogeneous data. The biomedical informatics group of the Universidad Politécnica de Madrid cooperates with the European project P-medicine which main purpose is to create an infrastructure and models to facilitate the transition from current medical practice to personalized medicine. One of the tasks of the project that the group is in charge of consists on the development of tools that will help users in the process of integrating data from diverse sources. Some of the sources are biomedical public data bases from the NCBI platform (National Center for Biotechnology Information). One of the tools in which the group is currently working on for the integration of data sources is called the Ontology Annotator. In this tool there is a phase in which the user has to retrieve information from a public data base and select the relevant data contained in it manually. For automating the process of searching and selecting data on the one hand, there is an interest in automatically generating queries that guide towards the more precise results as possible. On the other hand, there is an interest on retrieve relevant information from large quantities of documents. The solution requires systems that analyze and weigh the data allowing the localization of the relevant items. In the computer science field of the artificial intelligence, in the branch of information retrieval there are diverse studies about the query expansion from relevance feedback that could be used to solve the problem. The main purpose of this studies is to obtain a set of results that is the closer as possible to the information that the user really wants to retrieve. In order to reach this purpose different techniques are used to reformulate or expand the initial query using a feedback the results that where relevant for the user, with this method, the new set of results will have more proximity with the ones that the user really desires. The goal of this final dissertation project consists on the study, implementation and experimentation of methods that automate the process of extraction of relevant information from documents using this information to expand queries. This way, the precision and the ranking of the results associated will be improved. These methods will be integrated in the Ontology Annotator tool and will focus on the PubMed data source.

A four-factor user interaction model for content-based image retrieval

In order to bridge the “Semantic gap”, a number of relevance feedback (RF) mechanisms have been applied to content-based image retrieval (CBIR). However current RF techniques in most existing CBIR systems still lack satisfactory user interaction although some work has been done to improve the interaction as well as the search accuracy. In this paper, we propose a four-factor user interaction model and investigate its effects on CBIR by an empirical evaluation. Whilst the model was developed for our research purposes, we believe the model could be adapted to any content-based search system.

Enabling effective user interactions in content-based image retrieval

This paper presents an interactive content-based image retrieval framework—uInteract, for delivering a novel four-factor user interaction model visually. The four-factor user interaction model is an interactive relevance feedback mechanism that we proposed, aiming to improve the interaction between users and the CBIR system and in turn users overall search experience. In this paper, we present how the framework is developed to deliver the four-factor user interaction model, and how the visual interface is designed to support user interaction activities. From our preliminary user evaluation result on the ease of use and usefulness of the proposed framework, we have learnt what the users like about the framework and the aspects we could improve in future studies. Whilst the framework is developed for our research purposes, we believe the functionalities could be adapted to any content-based image search framework.

InfoScout: An interactive, entity centric, person search tool.

Individuals living in highly networked societies publish a large amount of personal, and potentially sensitive, information online. Web investigators can exploit such information for a variety of purposes, such as in background vetting and fraud detection. However, such investigations require a large number of expensive man hours and human effort. This paper describes InfoScout, a search tool which is intended to reduce the time it takes to identify and gather subject centric information on the Web. InfoScout collects relevance feedback information from the investigator in order to rerank search results, allowing the intended information to be discovered more quickly. Users may still direct their search as they see fit, issuing ad-hoc queries and filtering existing results by keywords. Design choices are informed by prior work and industry collaboration.

Learning Binary Code Representations for Effective and Efficient Image Retrieval

The size of online image datasets is constantly increasing. Considering an image dataset with millions of images, image retrieval becomes a seemingly intractable problem for exhaustive similarity search algorithms. Hashing methods, which encodes high-dimensional descriptors into compact binary strings, have become very popular because of their high efficiency in search and storage capacity. In the first part, we propose a multimodal retrieval method based on latent feature models. The procedure consists of a nonparametric Bayesian framework for learning underlying semantically meaningful abstract features in a multimodal dataset, a probabilistic retrieval model that allows cross-modal queries and an extension model for relevance feedback. In the second part, we focus on supervised hashing with kernels. We describe a flexible hashing procedure that treats binary codes and pairwise semantic similarity as latent and observed variables, respectively, in a probabilistic model based on Gaussian processes for binary classification. We present a scalable inference algorithm with the sparse pseudo-input Gaussian process (SPGP) model and distributed computing. In the last part, we define an incremental hashing strategy for dynamic databases where new images are added to the databases frequently. The method is based on a two-stage classification framework using binary and multi-class SVMs. The proposed method also enforces balance in binary codes by an imbalance penalty to obtain higher quality binary codes. We learn hash functions by an efficient algorithm where the NP-hard problem of finding optimal binary codes is solved via cyclic coordinate descent and SVMs are trained in a parallelized incremental manner. For modifications like adding images from an unseen class, we propose an incremental procedure for effective and efficient updates to the previous hash functions. Experiments on three large-scale image datasets demonstrate that the incremental strategy is capable of efficiently updating hash functions to the same retrieval performance as hashing from scratch.

Relevance feature discovery for text analysis

It is a big challenge to guarantee the quality of discovered relevance features in text documents for describing user preferences because of the large number of terms, patterns, and noise. Most existing popular text mining and classification methods have adopted term-based approaches. However, they have all suffered from the problems of polysemy and synonymy. Over the years, people have often held the hypothesis that pattern-based methods should perform better than term- based ones in describing user preferences, but many experiments do not support this hypothesis. This research presents a promising method, Relevance Feature Discovery (RFD), for solving this challenging issue. It discovers both positive and negative patterns in text documents as high-level features in order to accurately weight low-level features (terms) based on their specificity and their distributions in the high-level features. The thesis also introduces an adaptive model (called ARFD) to enhance the exibility of using RFD in adaptive environment. ARFD automatically updates the system's knowledge based on a sliding window over new incoming feedback documents. It can efficiently decide which incoming documents can bring in new knowledge into the system. Substantial experiments using the proposed models on Reuters Corpus Volume 1 and TREC topics show that the proposed models significantly outperform both the state-of-the-art term-based methods underpinned by Okapi BM25, Rocchio or Support Vector Machine and other pattern-based methods.

Role of feedback in mammalian vision: a new hypothesis and a computational model

This paper presents a novel hypothesis on the function of massive feedback pathways in mammalian visual systems. We propose that the cortical feature detectors compete not for the right to represent the output at a point, but for exclusive rights to abstract and represent part of the underlying input. Feedback can do this very naturally. A computational model that implements the above idea for the problem of line detection is presented and based on that we suggest a functional role for the thalamo-cortical loop during perception of lines. We show that the model successfully tackles the so called Cross problem. Based on some recent experimental results, we discuss the biological plausibility of our model. We also comment on the relevance of our hypothesis (on the role of feedback) to general sensory information processing and recognition. (C) 1998 Published by Elsevier Science Ltd. All rights reserved.

Feedback von Schülerinnen und Schülern als Quelle des Lernprozesses im Praktikum von Lehramtsstudierenden

Rückmeldungen von Schülerinnen und Schülern zum Unterricht nehmen eine wichtige Rolle zur Verbesserung der Unterrichtsqualität ein. Während der Ausbildung von Lehrpersonen wird dieses Potenzial bisher aber selten genutzt. Der Beitrag stellt Ergebnisse aus dem Forschungsprojekt "Lernen im Praktikum" vor. Befragt wurden 150 Studierende des Lehramts für 7.-9. Klassen (Sekundarstufe 1) zur Bedeutung von Schülerrückmeldungen im Praktikum sowie 1331 Schülerinnen und Schüler aus Praktikumsklassen zum Praktikum und zum Lernprozess der Praktikantinnen und Praktikanten. Die Ergebnisse unterstützen einerseits die Relevanz systematischer Schülerrückmeldungen für das Lernen im Praktikum. Andererseits machen sie auf die Notwendigkeit aufmerksam, die Kriterien, die Schülerbeurteilungen zugrunde liegen, weiter zu erforschen.