Automatic Information Retrieval Research Articles

Arabic Dialect Identification based on Probabilistic-Phonetic Modeling

The identification of Arabic dialects is considered to be the first pre-processing component for any natural language processing problem. This task is useful for automatic translation, information retrieval, opinion mining and sentiment analysis. In this purpose, we propose a statistical approach based on the phonetic modeling to identify the correspondent Arabic dialect for each input acoustic signal. The main idea consists first, and for each dialect, in calculating a referenced phonetic model. Second, for every input audio signal, we calculate an appropriate phonetic model. Third, we compare this latter to all referenced Arabic dialect models. Finally, we associate the input acoustic signal to the dialect where the referenced phonetic model minimizes the cosine similarity. The obtained results are satisfactory. Indeed, based on 117 audio sequences, we attain a classification rate of 93%. Supporting the achieved results and the coverage of most of Arabic dialects, this study can be a reference for future work addressing dialectical speech processing applications.

A selectional auto-encoder approach for document image binarization

Binarization plays a key role in the automatic information retrieval from document images. This process is usually performed in the first stages of document analysis systems, and serves as a basis for subsequent steps. Hence it has to be robust in order to allow the full analysis workflow to be successful. Several methods for document image binarization have been proposed so far, most of which are based on hand-crafted image processing strategies. Recently, Convolutional Neural Networks have shown an amazing performance in many disparate duties related to computer vision. In this paper we discuss the use of convolutional auto-encoders devoted to learning an end-to-end map from an input image to its selectional output, in which activations indicate the likelihood of pixels to be either foreground or background. Once trained, documents can therefore be binarized by parsing them through the model and applying a global threshold. This approach has proven to outperform existing binarization strategies in a number of document types.

Coherence in general and personal semantic knowledge: functional differences of the posterior and centro-parietal N400 ERP component.

A growing body of scientific literature investigated the difference between general and personal semantic knowledge. In contrast to general world knowledge, personal semantics comprises highly individual knowledge about oneself. The present study aimed to differentiate processes of integration into personal as opposed to general semantic knowledge. For that purpose, participants were presented with pictures of themselves (Self-condition) or unknown persons (Other-condition) superimposed on a congruent or incongruent background. We hypothesized that self-referential processing is based on automatic retrieval of personal information as opposed to the processing of unknown persons, which requires voluntary, i.e., strategic, attention demanding processing. The topography of the N400 effect varied as a function of the type of semantic knowledge. We found a centro-parietal N400 effect within the Self-condition and a posterior effect within the Other-condition. The voluntary integration of facial expressions of unknown persons within the Other-condition was, furthermore, indexed by an N170 effect. The unresolved tension in personal semantics was reflected by the N500. Our study thus provides new impulses for interpretation of the N400's functional properties and extends our knowledge about the N500. Implications for the functional properties of the self as an organizational structure are discussed.

Spoken keyword search system using improved ASR engine and novel template-based keyword scoring

Keyword search for spoken documents has become more and more important nowadays due to the increasing amount of spoken data. The typical system makes use of an Automatic Speech Recognition system (ASR) and information retrieval methods. While a number of studies have been done to get the optimal system performance, KeyWord Search (KWS) systems still suffer from two main drawbacks. First, the system performance depends strongly on the ASR transcripts which are inherently inexact. Due to the speech signal variabilities, ASR systems are far from being powerful. Second, KWS systems make detection decisions based on the lattice-based posterior probability which is incomparable across keywords. In addition, posterior probabilities of true detection usually fall into different ranges which decrease the spotting performance. This paper considers the problems of ASR transcriptions and keyword detection decision based on posterior probabilities. More specifically, we propose to enhance the ASR transcripts accuracy by introducing a new ASR architecture in which we integrate data augmentation and ensemble learning techniques into a single framework. In addition, we proposed a novel keyword rescoring method that provides scores from a new perspective. Precisely, inspired by template-based KWS approach, scores of similarity between the detected keywords are computed by computing the distance between the acoustic features and are used as new scores for decision. Experiments on French and English datasets show that the proposed KWS system potentially leads to more accurate keyword results than the conventional systems.

Rumination impairs the control of stimulus-induced retrieval of irrelevant information, but not attention, control, or response selection in general

The aim of the study was to throw more light on the relationship between rumination and cognitive-control processes. Seventy-eight adults were assessed with respect to rumination tendencies by means of the LEIDS-r before performing a Stroop task, an event-file task assessing the automatic retrieval of irrelevant information, an attentional set-shifting task, and the Attentional Network Task, which provided scores for alerting, orienting, and executive control functioning. The size of the Stroop effect and irrelevant retrieval in the event-five task were positively correlated with the tendency to ruminate, while all other scores did not correlate with any rumination scale. Controlling for depressive tendencies eliminated the Stroop-related finding (an observation that may account for previous failures to replicate), but not the event-file finding. Taken altogether, our results suggest that rumination does not affect attention, executive control, or response selection in general, but rather selectively impairs the control of stimulus-induced retrieval of irrelevant information.

Similarity‐Based Summarization of Music Files for Support Vector Machines

Automatic retrieval of music information is an active area of research in which problems such as automatically assigning genres or descriptors of emotional content to music emerge. Recent advancements in the area rely on the use of deep learning, which allows researchers to operate on a low‐level description of the music. Deep neural network architectures can learn to build feature representations that summarize music files from data itself, rather than expert knowledge. In this paper, a novel approach to applying feature learning in combination with support vector machines to musical data is presented. A spectrogram of the music file, which is too complex to be processed by SVM, is first reduced to a compact representation by a recurrent neural network. An adjustment to loss function of the network is proposed so that the network learns to build a representation space that replicates a certain notion of similarity between annotations, rather than to explicitly make predictions. We evaluate the approach on five datasets, focusing on emotion recognition and complementing it with genre classification. In experiments, the proposed loss function adjustment is shown to improve results in classification and regression tasks, but only when the learned similarity notion corresponds to a kernel function employed within the SVM. These results suggest that adjusting deep learning methods to build data representations that target a specific classifier or regressor can open up new perspectives for the use of standard machine learning methods in music domain.

Бібліографічна база даних як джерело наукометричних досліджень

It is stated in the article that one of the most important characteristics of the quality of a scientific publication is its bibliographic apparatus, and these are bibliographic references, books and reference lists. The author has found out that bibliographic information makes it possible to determine the source of the research, show the relationship of the new publication with the previous ones, make an idea of the scientific position of the author, verify the accuracy of the data, disclose the priority and scientific value of the results, protect copyright, fight plagiarism. It has been proved that in the evaluation of the effectiveness of scientific activity, scientometrics play an important role – the area of the research which studies cognitive communication in science by the frequency of references and citations of scientific publications and their authors. Based on the analysis of domestic and foreign publications, the issue related to the bibliographic support of scientometric studies of literature for scientific work has been highlighted, its relevance and insufficient development have been substantiated and the main approaches to the elimination of “white spots” for compiling books and reference lists have been outlined. It has been emphasized that books and article lists would be even more efficient if they were compiled at a higher level and standardized bibliographic records were used. This would make it possible to use these lists in the help desks of various information institutions and libraries, to enter them into the databases of automated information retrieval systems without additional verification of the accuracy of the indicated information. It has been proved that the bibliographic list for a scientific work helps to increase the competence and skills of the authors in selecting and analysing the source research base, unifying the bibliographic description of documents of various types, and makes it easier for authors to work with standards and other normative and methodological documents for the bibliographic database of their scientific works.

Generating a Tolerogenic Cell Therapy Knowledge Graph from Literature.

Tolerogenic cell therapies provide an alternative to conventional immunosuppressive treatments of autoimmune disease and address, among other goals, the rejection of organ or stem cell transplants. Since various methodologies can be followed to develop tolerogenic therapies, it is important to be aware and up to date on all available studies that may be relevant to their improvement. Recently, knowledge graphs have been proposed to link various sources of information, using text mining techniques. Knowledge graphs facilitate the automatic retrieval of information about the topics represented in the graph. The objective of this work was to automatically generate a knowledge graph for tolerogenic cell therapy from biomedical literature. We developed a system, ICRel, based on machine learning to extract relations between cells and cytokines from abstracts. Our system retrieves related documents from PubMed, annotates each abstract with cell and cytokine named entities, generates the possible combinations of cell–cytokine pairs cooccurring in the same sentence, and identifies meaningful relations between cells and cytokines. The extracted relations were used to generate a knowledge graph, where each edge was supported by one or more documents. We obtained a graph containing 647 cell–cytokine relations, based on 3,264 abstracts. The modules of ICRel were evaluated with cross-validation and manual evaluation of the relations extracted. The relation extraction module obtained an F-measure of 0.789 in a reference database, while the manual evaluation obtained an accuracy of 0.615. Even though the knowledge graph is based on information that was already published in other articles about immunology, the system we present is more efficient than the laborious task of manually reading all the literature to find indirect or implicit relations. The ICRel graph will help experts identify implicit relations that may not be evident in published studies.

Pivoted Document Length Normalization

Automatic information retrieval systems have to deal with documents of varying lengths in a text collection. Document length normalization is used to fairly retrieve documents of all lengths. In this study, we ohserve that a normalization scheme that retrieves documents of all lengths with similar chances as their likelihood of relevance will outperform another scheme which retrieves documents with chances very different from their likelihood of relevance. We show that the retrievaf probabilities for a particular normalization method deviate systematically from the relevance probabilities across different collections. We present pivoted normalization, a technique that can be used to modify any normalization function thereby reducing the gap between the relevance and the retrieval probabilities. Training pivoted normalization on one collection, we can successfully use it on other (new) text collections, yielding a robust, collectzorz independent normalization technique. We use the idea of pivoting with the well known cosine normalization function. We point out some shortcomings of the cosine function andpresent two new normalization functions--pivoted unique normalization and piuotert byte size normalization.

Identifying New Product Ideas: Waiting for the Wisdom of the Crowd or Screening Ideas in Real Time

Crowdsourcing ideas from consumers can enrich idea input in new product development. After a decade of initiatives (e.g., Starbucks’ MyStarbucksIdea, Dell's IdeaStorm), the implications of crowdsourcing for idea generation are well understood, but challenges remain in dealing with the large volume of rapidly generated ideas produced in crowdsourcing communities. This study proposes a model that can assist managers in efficiently processing crowdsourced ideas by identifying the aspects of ideas that are most predictive of future implementation and identifies three sources of information available for an idea: its content, the contributor proposing it, and the crowd's feedback on the idea (the “3Cs”). These information sources differ in their time of availability (content/contributor information is available immediately; crowd feedback accumulates over time) and in the extent to which they comprise structured or unstructured data. This study draws from prior research to operationalize variables corresponding to the 3Cs and develops a new measure to quantify an idea's distinctiveness. Applying automated information retrieval methods (latent semantic indexing) and testing several linear methods (linear discriminant analysis, regularized logistic regression) and nonlinear machine‐learning algorithms (stochastic adaptive boosting, random forests), this article identifies the variables that are most useful towards predicting idea implementation in a crowdsourcing community for an IT product (Mendeley). Our results indicate that consideration of content and contributor information improves ranking performance between 22.6 and 26.0% over random idea selection, and that adding crowd‐related information further improves performance by up to 48.1%. Crowd feedback is the best predictor of idea implementation, followed by idea content and distinctiveness, and the contributor's past idea‐generation experience. Firms are advised to implement two idea selection support systems: one to rank new ideas in real time based on content and contributor experience, and another that integrates the crowd's idea evaluation after it has had sufficient time to provide feedback.

Verb generation task: early automatic information retrieval or late effortful decision-making?

Verb generation task: early automatic information retrieval or late effortful decision-making?

Scattering-Based Nonlocal Means SAR Despeckling

Speckle noise greatly limits both synthetic aperture radar (SAR) data human readability, especially for non-SAR-expert users, and performance of automatic processing and information retrieval procedures by computer programs. Therefore, despeckling of SAR images is an essential preprocessing step in SAR data analysis, processing, and modeling, as well as in information retrieval and inversion procedures. Up to now, one of the most accurate and promising despeckling approaches—among those based on a single SAR image—is the one relying on the nonlocal means concepts. However, at the best of our knowledge, most of the state of the art considers the despeckling problem only within a statistical framework, completely discarding the electromagnetic phenomena behind SAR imagery formation. In this paper, we introduce the novel idea of a physical-based despeckling, taking into account meaningful physical characteristics of the imaged scenes. This idea is realized via the implementation of a physical-oriented probabilistic patch-based (PPB) filter based on a priori knowledge of the underlying topography and analytical scattering models. This filter is suitable for SAR images of natural scenes presenting a significant topography. An adaptive version of the proposed scattering-based PPB filter for denoising of SAR images including both mountainous and flat areas is also developed. The performances of the proposed filter and its adaptive version are evaluated both qualitatively and quantitatively in numerical experiments using both simulated and actual SAR images. The proposed technique exhibits performance superior w.r.t. the standard PPB filter and comparable or, in some cases, superior to the state of the art, both in terms of speckle reduction and texture and detail preservation.

Digitising legacy zoological taxonomic literature: Processes, products and using the output

By digitising legacy taxonomic literature using XML mark-up the contents become accessible to other taxonomic and nomenclatural information systems. Appropriate schemas need to be interoperable with other sectorial schemas, atomise to appropriate content elements and carry appropriate metadata to, for example, enable algorithmic assessment of availability of a name under the Code. Legacy (and new) literature delivered in this fashion will become part of a global taxonomic resource from which users can extract tailored content to meet their particular needs, be they nomenclatural, taxonomic, faunistic or other.To date, most digitisation of taxonomic literature has led to a more or less simple digital copy of a paper original – the output of the many efforts has effectively been an electronic copy of a traditional library. While this has increased accessibility of publications through internet access, the means by which many scientific papers are indexed and located is much the same as with traditional libraries. OCR and born-digital papers allow use of web search engines to locate instances of taxon names and other terms, but OCR efficiency in recognising taxonomic names is still relatively poor, people’s ability to use search engines effectively is mixed, and many papers cannot be searched directly. Instead of building digital analogues of traditional publications, we should consider what properties we require of future taxonomic information access. Ideally the content of each new digital publication should be accessible in the context of all previous published data, and the user able to retrieve nomenclatural, taxonomic and other data / information in the form required without having to scan all of the original papers and extract target content manually. This opens the door to dynamic linking of new content with extant systems: automatic population and updating of taxonomic catalogues, ZooBank and faunal lists, all descriptions of a taxon and its children instantly accessible with a single search, comparison of classifications used in different publications, and so on. A means to do this is through marking up content into XML, and the more atomised the mark-up the greater the possibilities for data retrieval and integration. Mark-up requires XML that accommodates the required content elements and is interoperable with other XML schemas, and there are now several written to do this, particularly TaxPub, taxonX and taXMLit, the last of these being the most atomised. We now need to automate this process as far as possible. Manual and automatic data and information retrieval is demonstrated by projects such as INOTAXA and Plazi. As we move to creating and using taxonomic products through the power of the internet, we need to ensure the output, while satisfying in its production the requirements of the Code, is fit for purpose in the future.

Codified Hashtags for Weather Warning on Twitter: an Italian Case Study

Introduction: During emergencies increasing numbers of messages are shared through social media platforms becoming a primary source of information for lay people and emergency managers. For Twitter codified hashtagging is emerging as a practical way to coordinate messages during emergencies and quickly identify relevant information. This paper considers a case study on the use of codified hashtags concerning weather warning in Italy in three different regions.Methods: From November 3rd to December 2nd 2014, tweets identified by the 3 codified hashtags #allertameteoTOS, #allertameteoLIG and #allertameteoPIE were retrieved, collecting a total of 35,558 tweets published by 7361 unique tweets authors, with the aim to assess if codified hashtags could represent an effective way to align formal and informal sources of information during weather related emergencies. An auxiliary R-package was built to lead the analytics used in this study. Authors performed a manual coding of users, hashtags and content of messages of all Twitter data considered.Results: Content analysis showed that tweets were overwhelmingly related to situational updates, with a high percentage containing geo-location information. Communication patterns of different user types were discussed for the three contexts. In accordance with previous studies, individuals showed an active participation primarily functioning as information hub during the emergency.Discussion: In the proposed cases codified hashtags have proven to be an effective tool to convey useful information on Twitter by formal and informal sources. Where institutions supported the use of the predefined hashtag in communication activities, like in Tuscany, messages were very focused, with more than 90% of tweets being situational updates. In this perspective, use of codified hashtags may potentially improve the performance of systems for automatic information retrieval and processing during disasters.Keywords: social media, emergency management, Twitter, severe weather

Automatic and Controlled Semantic Retrieval: TMS Reveals Distinct Contributions of Posterior Middle Temporal Gyrus and Angular Gyrus.

We demonstrate a novel functional dissociation between the angular gyrus (AG) and posterior middle temporal gyrus (pMTG) in conceptual processing. These sites are often coactivated during neuroimaging studies using semantic tasks, but their individual contributions are unclear. Using transcranial magnetic stimulation and tasks designed to assess different aspects of semantics (item identity and thematic matching), we tested two alternative theoretical accounts. Neither site showed the pattern expected for a "thematic hub" (i.e., a site storing associations between concepts) since stimulation disrupted both tasks. Instead, the data indicated that pMTG contributes to the controlled retrieval of conceptual knowledge, while AG is critical for the efficient automatic retrieval of specific semantic information.

What a Computer-Based Legal Reference Work Can and Must Deliver

This paper was presented at a conference on electronic publishing sponsored by the National Center for Automated Information Retrieval in 1991. It draws on the author's work designing and building an electronic treatise covering Social Security law, within the limitations of the LEXIS online service of the period. It reviews the more obvious advantages works of this kind can offer before focusing on the very different relationship between author text and authoritative primary material possible in this new research environment.

Ten years of science news: A longitudinal analysis of scientific culture in the Spanish digital press.

This article presents our study of science coverage in the digital Spanish press over the last decade. We employed automated information retrieval procedures to create a corpus of 50,763 text units dealing with science and technology, and used automated text-analysis procedures in order to provide a general picture of the structure, characteristics and evolution of science news in Spain. We found between 6% and 7% of science coverage, a clear high proportion of biomedicine and predominance of science over technology, although we also detected an increase in technological content during the second half of the decade. Analysing the extrinsic and intrinsic features of science culture, we found a predominance of intrinsic features that still need further analysis. Our attempt to use specialised software to examine big data was effective, and allowed us to reach these preliminary conclusions.

GA-Based Adaptive Window Length Estimation for Highly Accurate Audio Segmentation

Accurate audio segmentation has recently received increasing attention for its applications in automatic indexing, content analysis and information retrieval. Hence, this paper proposes a highly accurate audio segmentation methodology using a genetic algorithm-based approach to adapting and optimizing segmentation window lengths. Specifically, this paper analyzes the parameter sequence of the root-mean-square values of an input audio stream with optimal sliding window (or segmentation window) lengths found and adapted by a genetic algorithm. In addition, this paper determines whether an audio-cut occurs or not by utilizing the parameter sequences as inputs of a support vector machine. Experimental results indicate that the proposed approach achieves 100.00% and 98.69% in the average precision and recall rates of segmentation performance, respectively.

Methods of Processing and Retrieval of Information from Digital Particle Holograms and Their Application

We consider a holographic study of particles. The main attention is focused on retrieval of information from a digital hologram, namely, determination of the characteristics of a particle and its holographic image. To estimate the efficiency of this stage, we used quantitative criteria of the holographic image quality. These criteria were also used for determining the position of a particle in space, which ensures a significantly higher accuracy (an error of about 0.2 mm) compared with other methods. An algorithm for increasing the quality of the holographic image of a particle is developed. A method for two-dimensional holographic mapping of the volume of the medium with particles for automatic retrieval of information from the digital particle hologram is discussed.

Research of Drug Name Entity Recognition Based on Constructed Dictionary and Conditional Random Field

Drug name entity recognition (NER) is an important foundation of information extraction, automatic question answering, machine translation and information retrieval and other natural language processing technology based on the medical literature. This paper presents a method combined a constructed dictionary and conditional random field model to identify the drug entity. The proposed method has good performance in DDIExtraction 2013 evaluation corpus. //