Current Information Retrieval Systems Research Articles

A User-Centric Multi-Context Hybrid Reasoning Information Retrieval Model

Information Retrieval has been in existence since the 1940s and is impossible to do without. However, current information retrieval systems are known to be ineffective. The reason that seems to encompass this ineffectiveness is design-focused. Designers and developers of Information Retrieval Systems are known to be system focused rather than user-focused. They provide the same information to users even when they are in different contexts and have diverse preferences. Using a restaurant use case, we propose a conceptual user-centric multi-context hybrid reasoning Information Retrieval model to improve the accuracy of retrieved results.

An Innovative Information Retrieval Model Implementing Particle Swarm Optimization Technique

The Internet has enormous information and it is growing rapidly. The vast amount of data creates challenges in relation to effective Information Retrieval (IR). The scope of the Information Retrieval System (IRS) is to provide the most relevant data for user query from large datasets. However the current IR system fails to provide the hidden and up to date data. This paper focused on soft computing techniques to overcome the above mentioned issues. Particle Swarm Optimization (PSO) is used to compute the fitness function to optimize the retrieval result. PSO has an efficient capability in global search and the implementation is easy to develop. The implementation result of the present study is feasible, that improves the retrieval effect and the accuracy of hidden data retrieval.

Search and analytics using semantic annotations

Search and analytics using semantic annotations

Fairness and Transparency in Ranking

Ranking in Information Retrieval (IR) has been traditionally evaluated from the perspective of the relevance of search engine results to people searching for information, i.e., the extent to which the system provides "the right information, to the right people, in the right way, at the right time." However, people in current IR systems are not only the ones issuing search queries, but increasingly they are also the ones being searched. This raises several new problems in IR that have been addressed in recent research, particularly with respect to fairness/non-discrimination, accountability, and transparency. This is a summary of some these initial developments.

Visual content-based web page categorization with deep transfer learning and metric learning

The growing amounts of online multimedia content challenge the current search, recommendation and information retrieval systems. Information in the form of visual elements is highly valuable in a range of web mining tasks. However, the mining of these resources is a difficult task due to the complexity and variability of images, and the cost of collecting big enough datasets to successfully train accurate deep learning models. This paper proposes a novel framework for the categorization of web pages on the basis of their visual content. This is achieved by exploring the joint application of a transfer learning strategy and metric learning techniques to build a Deep Convolutional Neural Network (DCNN) for feature extraction, even when training data is scarce. The obtained experimental results evidence that the proposed approach outperforms the state-of-the-art handcrafted image descriptors and achieves a high categorization accuracy. In addition, we address the problem of over-time learning, so the proposed framework can learn to identify new web page categories as new labeled images are provided at test time. As a result, prior knowledge of the complete set of possible web categories is not necessary in the initial training phase.

Mixed language queries in online searches

PurposeWith the increasing number of online multilingual resources, cross-language information retrieval (CLIR) has drawn much attention from the information retrieval (IR) research community. However, few studies have examined how and why multilingual searchers seek information in two or more languages, specifically how they switch and mix language in queries to get satisfying results. The purpose of this paper is to focus on Chinese–English bilinguals’ intra-sentential code-switching behaviors in online searches. The scenarios and reasons of code-switching, factors that may affect code-switching, the patterns of mixed language query formulation and reformulation and how current IR systems and other search tools can facilitate such information needs were examined.Design/methodology/approachIn-depth semi-structured interviews were used as the research method. In total, 30 participants were recruited based on their English proficiency, location and profession, using a purposive sampling method.FindingsFour scenarios and four reasons for using Chinese–English mixed language queries to cover information needs were identified, and results suggest that linguistic and cultural/social factors are of equivalent importance in code-switching behaviors. English terms and Chinese terms in queries play different roles in searches, and mixed language queries are irreplaceable by either single language queries or other search facilitating features. Findings also suggest current search engines and tools need greater emphasis in the user interface and more user education is required.Originality/valueThis study presents a qualitative analysis of bilinguals’ code-switching behaviors in online searches. Findings are expected to advance the theoretical understanding of bilingual users’ search strategies and interactions with IR systems, and provide insights for designing more effective IR systems and tools to discover multilingual online resources, including cross-language controlled vocabularies, personalized CLIR tools and mixed language query assistants.

Searching bibliographic data using graphs: A visual graph query interface

Abstract With the ever-increasing scientific literature, improving the efficiency of searching bibliographic data has become an important issue. With a lack of support of current bibliographic information retrieval systems in expressing complicated information needs, getting relevant bibliographic data is a demanding task. In this paper, we propose a visual graph query interface for bibliographic information retrieval. Through this interface, users can formulate bibliographic queries by interacting with a graph. Visual graph queries use a set of nodes with constraints and links among nodes to represent explicit and precise bibliographic information needs. The proposed visual graph query interface allows users to formulate several complex bibliographic queries (e.g., bibliographic coupling) that are not attainable in current major bibliographic information retrieval systems. In addition, the proposed interface requires less number of queries in completing everyday bibliographic search tasks.

Design implications for task-specific search utilities for retrieval and re-engineering of code

ABSTRACTThe importance of information retrieval systems is unquestionable in the modern society and both individuals as well as enterprises recognise the benefits of being able to find information effectively. Current code-focused information retrieval systems such as Google Code Search, Codeplex or Koders produce results based on specific keywords. However, these systems do not take into account developers’ context such as development language, technology framework, goal of the project, project complexity and developer’s domain expertise. They also impose additional cognitive burden on users in switching between different interfaces and clicking through to find the relevant code. Hence, they are not used by software developers. In this paper, we discuss how software engineers interact with information and general-purpose information retrieval systems (e.g. Google, Yahoo!) and investigate to what extent domain-specific search and recommendation utilities can be developed in order to support their work-related activities. In order to investigate this, we conducted a user study and found that software engineers followed many identifiable and repeatable work tasks and behaviours. These behaviours can be used to develop implicit relevance feedback-based systems based on the observed retention actions. Moreover, we discuss the implications for the development of task-specific search and collaborative recommendation utilities embedded with the Google standard search engine and Microsoft IntelliSense for retrieval and re-engineering of code. Based on implicit relevance feedback, we have implemented a prototype of the proposed collaborative recommendation system, which was evaluated in a controlled environment simulating the real-world situation of professional software engineers. The evaluation has achieved promising initial results on the precision and recall performance of the system.

Evaluation of semantic similarity metrics applied to the automatic retrieval of medical documents: An UMLS approach

One promise of current information retrieval systems is the capability to identify risk groups for certain diseases and pathologies based on the automatic analysis of vast amounts of Electronic Medical Records repositories. However, the complexity and the degree of specialization of the language used by the experts in this context, make this task both challenging and complex. In this work, we introduce a novel experimental study to evaluate the performance of the two semantic similarity metrics (Path and Intrinsic IC-Path, both widely accepted in the literature) in a real-life information retrieval situation. In order to achieve this goal and due to the lack of methodologies for this context in the literature, we propose a straightforward information retrieval system for the biomedical field based on the UMLS Metathesaurus and on semantic similarity metrics. In contrast with previous studies which focus on testbeds with limited and controlled sets of concepts, we use a large amount of information (101,712 medical documents extracted from TREC Medical Records Track 2011). Our results show that in real-life cases, both metrics display similar performance, Path (F-Measure = 0.430) e Intrinsic IC-Path (F-Measure = 0.427). Thereby we suggest that the use of Intrinsic IC-Path is not justified in real scenarios.

A library’s information retrieval system (In)effectiveness: case study

Purpose– The purpose of this paper is to evaluate the effectiveness of the information retrieval component of a daily newspaper publisher’s integrated library system (ILS) in comparison with the open source alternatives and observe the impact of the scale of metadata, generated daily by library administrators, on retrieved result sets.Design/methodology/approach– In Experiment 1, the authors compared the result sets of the information retrieval system (IRS) component of the publisher’s current ILS and the result sets of proposed ones with human-assessed relevance judgment set. In Experiment 2, the authors compared the performance of proposed IRS components with the publisher’s current production IRS, using result sets of current IRS classified as relevant. Both experiments were conducted using standard information retrieval (IR) evaluation methods: precision, recall, precision atk,F-measure, mean average precision and 11-point interpolated average precision.Findings– Results showed that: first, in Experiment 1, the publisher’s current production ILS ranked last of all participating IRSs when compared to a relevance document set classified by the senior library administrator; and second, in Experiment 2, the tested IR components’ request handlers that used only automatically generated metadata performed slightly better than request handlers that used all of the metadata fields. Therefore, regarding the effectiveness of IR, the daily human effort of generating the publisher’s current set of metadata attributes is unjustified.Research limitations/implications– The experiments’ collections contained Slovene language with large number of variations of the forms of nouns, verbs and adjectives. The results could be different if the experiments’ collections contained languages with different grammatical properties.Practical implications– The authors have confirmed, using standard IR methods, that the IR component used in the publisher’s current ILS, could be adequately replaced with an open source component. Based on the research, the publisher could incorporate the suggested open source IR components in practice. In the research, the authors have described the methods that can be used by libraries for evaluating the effectiveness of the IR of their ILSs.Originality/value– The paper provides a framework for the evaluation of an ILS’s IR effectiveness for libraries. Based on the evaluation results, the libraries could replace the IR components if their current information system setup allows it.

Modeling of food technology knowledge base information system using semantic web technologies

The impact of knowledge explosion in various fields during the recent years has resulted in creation of vast amount of scientific literature. Food Science &Technology (FST) is also an important subject domain where rapid developments are taking place due to large and diverse research and development activities. Due to these, information storage and retrieval has become very complex and current information retrieval systems (IRs) are constrained in achieving a high precision with good response time. To overcome these constraints as well as provide natural language based effective context based retrieval mechanism, a knowledge engineering framework need to be applied to represent, share and discover information. Semantic web technologies provides the mechanism for creating knowledge based systems, ontologies and rules for handling the data, ultimately results in intelligent information retrieval. Ontologies are the backbone of these types of knowledge systems. Ontology is a knowledge represented on the basis of conceptualization that intends a description of object and concept sets and relations between them. The set of objects, and the description of relationship among them, reflected the representational vocabulary with which a knowledge-based program knowledge.

A Novel Data Centric Information Retrieval Protocol for Queries in Delay Tolerant Networks

Information Retrieval (IR) systems aim to retrieve data that satisfies certain requirements and constitute an important service in many types of networks, including Delay/Disruption Tolerant Networks (DTNs). In current DTN based IR systems, the data that satisfies a query is assumed to be stored on a single node. Therefore, once a node receives a query in which it has the corresponding data, the query can be resolved completely. However, in scenarios where a query requires data from multiple nodes, these IR systems may fail. Henceforth, in this paper, we propose Distributed Data-Centric Information Retrieval (DDC-IR), a data centric IR system that supports all query types; e.g., continuous and complex. More importantly, it is designed specifically to operate in DTNs. It also incorporates a new packet, aka Query Reply Packet, that includes both a query and one or more replies. We show how this packet facilitates efficient query resolution and enables data centric routing. In addition, it uses caching so that nodes store popular queries that has the effect of speeding up query resolution. We have conducted an extensive simulation study to compare DDC-IR to state of the art IR systems using the popular Random Waypoint model and a trace-file containing student movements on a campus. The results show that DDC-IR is able to resolve 50 % more queries and has an 80 % lower buffer occupancy level than existing IR systems. We also tested DDC-IR in networks with varying sizes. For networks with 100 nodes, DDC-IR is able to resolve queries while current IR systems fail to resolve any queries. In particular, when the number of nodes increases, current IR systems fail to resolve any queries, whilst DDC-IR is able to resolve complex and continuous queries. The influence of the number of sub-queries on query resolution time is also studied. Specifically, when the number of sub-queries in a complex query increases from five to nine, DDC-IR uses 50 % more time to resolve a query. In comparison, prior IR systems fail to resolve any queries.

An effective query recommendation approach using semantic strategies for intelligent information retrieval

With the explosive growth of web information, search engines have become the mainstream tools of information retrieval (IR). However, a notable problem emerged in the current IR systems is that the input queries are usually too short and too ambiguous to express their actual idea which largely affects the performance of IR systems. In this study, a novel query recommendation technology which suggests a list of related queries is proposed to resolve these problems. The query concepts can be firstly extracted from the web-snippets of the search result returned by the input query. A bipartite graph is subsequently built to identify the related queries, and the query similarity can be calculated by such bipartite graph. Moreover, by analyzing the URLs clicked by users, we find that some tokens appeared in URLs are very meaningful, especial for some typical topic-based pages. Therefore, these potential tokens which can provide a brief description from the subject of the URL are also considered. In order to reveal the real semantics between queries, the approach TF-IQF model is further discussed, and three features of a query, i.e. clicked documents, associated query and reversed query, are utilized in our approach in depth. Such a method could hopefully acquire the comprehensive idea of a query. To investigate how these three features could be used effectively for query recommendation in search engine, we adopt the benchmark evaluation criterions in our experiments, and the experimental results show its promising results in comparison with state of the art methods.

An improved semantic information searching scheme based multi-agent system and an innovative similarity measure

The key task for the web, namely, web searches, is evolving towards some novel form of semantic web search. In fact, most information retrieval systems are based on static vectors representations. Two major difficulties when a researcher uses current information retrieval systems are how to filter out irrelevant documents, and how to discover latest or more significant documents. Recently a very promising approach to semantic web search is based on combining standard web pages and search queries with ontological background knowledge. In this perspective we will describe a model to hold a document’s noise, and incompleteness. For this, we merge a syntactic keyword search with purely semantic search based domain ontology and a multi-agent system to solve such distributed problems. Then we perform a ranking algorithm on returned documents, and we propose a new semantic similarity measure between concepts based on the WordNet taxonomy structure.

Context based Indexing in Search Engines using Ontology

Indexing in search engines has been an active area of current researches. The main aim of search engines is to provide most relevant documents to the users in minimum possible time. So granting efficient and fast accesses to the index is a major issue for performances of Web Search Engines. Indexing is performed on the web pages after they have been gathered into a repository by the crawler. The existing architecture of search engine shows that the index is built on the basis of the terms of the document and consists of an array of the posting lists where each posting list is associated with a term and contains the term as well as the identifiers of the documents containing the term. The current information retrieval systems use terms to describe documents and search engines. This paper proposes an indexing structure in which index is built on the basis of context of the document rather than on the terms basis using ontology. The ontology-based collection selection method presented in this paper uses context to describe collections and search engines. The context of the documents being collected by the crawler in the repository is being extracted by the indexer using the context repository, thesaurus and ontology repository and then documents are indexed according to their respective context. General Terms Algorithm, Theory, Performance.

A Hybrid Semantic Similarity Measure for Spatial Information Retrieval

Semantic similarity is central to many cognitive processes and plays an important role in the way humans process and reason about information. In particular, the retrieval of knowledge from memory hinges crucially on similarity. Likewise, information retrieval systems use similarity to detect relevant information for a given query. Current information retrieval systems apply mainly syntactic techniques to determine similarity. Although such syntactic similarity measures have performed strongly with resources containing large amounts of text, they cannot appropriately cope with syntactic and semantic heterogeneity and ambiguity, if the semantics of the terms is not explicitly available. Therefore, they are rather rigid and inflexible as they cannot adapt to the user's requirements and conceptualization of the domain. Furthermore, geographic features are distinguished via their geometric and thematic data. It is often not possible to capture the complex semantics of geographic features by a single name or a textual description. Therefore spatial data is different from text documents typically found in enterprise databases or on the Web. Retrieval of spatial information requires new, intelligent retrieval mechanisms that satisfy its specific requirements. A semantics-based solution can more easily adapt to user needs and therefore increases the flexibility and usability of spatial data and retrieval methods. This paper investigates the suitability of various approaches—originally developed to explain human similarity judgement—in the context of spatial information retrieval. We propose a new, hybrid approach for semantic similarity measurement, which can represent the complex semantics of spatial data. It allows for retrieving relevant data by determining the similarity between the query and the semantic descriptions of geographic feature types within the database. The hybrid similarity measure combines the geometric structure of conceptual spaces with the relational structure of semantic nets to one, cognitively plausible knowledge representation with an inherent similarity measure.

On the value of temporal information in information retrieval

Time is an important dimension of any information space and can be very useful in information retrieval. Current information retrieval systems and applications do not take advantage of all the time information available in the content of documents to provide better search results and user experience. In this paper we show some of the areas that can benefit from exploiting such temporal information.

Image retrieval: Benchmarking visual information indexing and retrieval systems

Image retrieval: Benchmarking visual information indexing and retrieval systems

A Bayesian Framework for XML Information Retrieval: Searching and Learning with the INEX Collection

Most recent document standards like XML rely on structured representations. On the other hand, current information retrieval systems have been developed for flat document representations and cannot be easily extended to cope with more complex document types. The design of such systems is still an open problem. We present a new model for structured document retrieval which allows computing scores of document parts. This model is based on Bayesian networks whose conditional probabilities are learnt from a labelled collection of structured documents--which is composed of documents, queries and their associated assessments. Training these models is a complex machine learning task and is not standard. This is the focus of the paper: we propose here to train the structured Bayesian Network model using a cross-entropy training criterion. Results are presented on the INEX corpus of XML documents.

Putting it together online: Information need identification for the domain novice user

AbstractDomain novice users in the beginning stages of researching a topic find themselves searching for information via information retrieval (IR) systems before they have identified their information need. Pre‐Internet access technologies adapted by current IR systems poorly serve these domain novice users, whose behavior might be characterized as rudderless and without a compass. In this article we describe a conceptual design for an information retrieval system that incorporates standard information need identification classification and subject cataloging schemes, called the INIIReye System, and a study that tests the efficacy of the innovative part of the INIIReye System, called the Associative Index. The Associative Index helps the user put together his or her associative thoughts—Vannevar Bush's idea of associative indexing for his Memex machine that he never actually described. For the first time, data from the study reported here quantitatively supports the theoretical notion that the information seeker's information need is identified through transformation of his/her knowledge structure (i.e., the seeker's cognitive map or perspective on the task for which information is being sought).