Information Retrieval Analysis Research Articles

Multiple sampling schemes and deep learning improve active learning performance in drug-drug interaction information retrieval analysis from the literature

BackgroundDrug-drug interaction (DDI) information retrieval (IR) is an important natural language process (NLP) task from the PubMed literature. For the first time, active learning (AL) is studied in DDI IR analysis. DDI IR analysis from PubMed abstracts faces the challenges of relatively small positive DDI samples among overwhelmingly large negative samples. Random negative sampling and positive sampling are purposely designed to improve the efficiency of AL analysis. The consistency of random negative sampling and positive sampling is shown in the paper.ResultsPubMed abstracts are divided into two pools. Screened pool contains all abstracts that pass the DDI keywords query in PubMed, while unscreened pool includes all the other abstracts. At a prespecified recall rate of 0.95, DDI IR analysis precision is evaluated and compared. In screened pool IR analysis using supporting vector machine (SVM), similarity sampling plus uncertainty sampling improves the precision over uncertainty sampling, from 0.89 to 0.92 respectively. In the unscreened pool IR analysis, the integrated random negative sampling, positive sampling, and similarity sampling improve the precision over uncertainty sampling along, from 0.72 to 0.81 respectively. When we change the SVM to a deep learning method, all sampling schemes consistently improve DDI AL analysis in both screened pool and unscreened pool. Deep learning has significant improvement of precision over SVM, 0.96 vs. 0.92 in screened pool, and 0.90 vs. 0.81 in the unscreened pool, respectively.ConclusionsBy integrating various sampling schemes and deep learning algorithms into AL, the DDI IR analysis from literature is significantly improved. The random negative sampling and positive sampling are highly effective methods in improving AL analysis where the positive and negative samples are extremely imbalanced.

Integrated Random Negative Sampling and Uncertainty Sampling in Active Learning Improve Clinical Drug Safety Drug–Drug Interaction Information Retrieval

Clinical drug–drug interactions (DDIs) have been a major cause for not only medical error but also adverse drug events (ADEs). The published literature on DDI clinical toxicity continues to grow significantly, and high-performance DDI information retrieval (IR) text mining methods are in high demand. The effectiveness of IR and its machine learning (ML) algorithm depends on the availability of a large amount of training and validation data that have been manually reviewed and annotated. In this study, we investigated how active learning (AL) might improve ML performance in clinical safety DDI IR analysis. We recognized that a direct application of AL would not address several primary challenges in DDI IR from the literature. For instance, the vast majority of abstracts in PubMed will be negative, existing positive and negative labeled samples do not represent the general sample distributions, and potentially biased samples may arise during uncertainty sampling in an AL algorithm. Therefore, we developed several novel sampling and ML schemes to improve AL performance in DDI IR analysis. In particular, random negative sampling was added as a part of AL since it has no expanse in the manual data label. We also used two ML algorithms in an AL process to differentiate random negative samples from manually labeled negative samples, and updated both the training and validation samples during the AL process to avoid or reduce biased sampling. Two supervised ML algorithms, support vector machine (SVM) and logistic regression (LR), were used to investigate the consistency of our proposed AL algorithm. Because the ultimate goal of clinical safety DDI IR is to retrieve all DDI toxicity–relevant abstracts, a recall rate of 0.99 was set in developing the AL methods. When we used our newly proposed AL method with SVM, the precision in differentiating the positive samples from manually labeled negative samples improved from 0.45 in the first round to 0.83 in the second round, and the precision in differentiating the positive samples from random negative samples improved from 0.70 to 0.82 in the first and second rounds, respectively. When our proposed AL method was used with LR, the improvements in precision followed a similar trend. However, the other AL algorithms tested did not show improved precision largely because of biased samples caused by the uncertainty sampling or differences between training and validation data sets.

Net ergonomics: information retrieval in health care domains.

The explosion of Internet-based socio-cultural subcontexts and the possibility, horizontally distributed, of handling information on the Net, are transforming information retrieval in a key topic for psychosocial research. However, describing and analyzing how a huge and complex topic can be discussed in the Net and how people's behaviors and attitudes can be influenced by this discussion, is a very complicated task. The aim of this chapter, is both to point out the limits of traditional approaches to the analysis of information retrieval in complex Internet-based domains, and to propose some advice for a cultural approach to the Net. In particular, starting from the analysis of a specific health care domain--drug abuse--the chapter will try to identify some guidelines for the study of these issues.

Cataloguing the Taxonomic Origins of Sequences from a Heterogeneous Sample Using Phylogenomics: Applications in Adventitious Agent Detection

We have designed and implemented a software system, named PhyloID™, that can be used to detect putative adventitious agents in biological samples characterized by next-generation sequencing. PhyloID is run in two steps, each being a self-contained automated process amenable to GMP validation. The first module, MiLY, is responsible for assembling individual sequence reads into contigs, and annotating all sequences with a unique sequence identifier, the number of reads in each contig, and the length of the sequence. The trimmed, assembled and annotated data are then processed by PhyloID's second module, NGmapper. NGmapper takes the FASTA-formatted output from MiLY and identifies the taxonomic origins of the contigs and singletons therein. It compares each sequence's BLASTN hit profile against the patterns of evolutionary relationships described within phylogenomic distance matrices for all of the various taxonomic groups, in order to find the best fit. NGmapper then produces lists of taxonomic assignments in both summarized and detailed form, and tree files for viewing results graphically. We optimized PhyloID's parameters and measured its performance using simulated metagenomic data and subsets of the reference phylogenies. PhyloID's precision and recall in identifying simulated sequences were measured by information retrieval analysis, focusing on read length, read number, sequence accuracy, background complexity, taxonomy and reference data coverage. We found PhyloID to be highly accurate and quantitative in its taxonomic mapping of sequences, with excellent precision, sensitivity and robustness. The degree of taxonomic representation available in publicly available databases remains an issue, as expected, for any sequence classifier, but community sequencing efforts are poised to overcome this problem. In order to illustrate real-world usage of the application, we also describe some simple spike-recovery experiments as well as a multi-site comparative characterization of a viral suspension. These data help to illustrate, to corroborate, and to extend results using simulated data. In order to address gaps in the detection of contaminating viruses and microorganisms in vaccines and other biologicals, manufacturers are exploring the use of new technologies that promise greater sensitivity and breadth of coverage. One challenge in implementing such new methods is the complexity of analysis of the "big data" generated by these new instruments: hundreds of millions of sequence reads (segments of genetic material from viruses and cells) need to be compared against a vast and growing number of entries in genetic databases, in order to come up with a confident identification. These large-scale analyses must furthermore be carried out within the strict regulatory environment that governs the industry. We have developed an automated software pipeline named PhyloID™ that is capable of identifying viruses and microorganisms from large-scale sequence data. Using simulated data as well as real samples, we show that PhyloID is both sensitive and accurate in identifying any type of potential contaminant. Such a powerful new assay will be an important addition to the adventitious agent testing package, providing further assurance about product safety.

User Profiling on a Pilot Digital Library with the Final Result of a New Adaptive Knowledge Management Solution

In this article, several procedures (e.g., measurements, information retrieval analyses, power law, association rules, hierarchical clustering) are introduced which were made on a pilot digital library. Information retrievals of web users from 01/01/2003 to 01/01/2006 on the internal search engine of the pilot digital library have been analyzed. With the power law method of data processing, a constant information retrieval pattern has been established, stable over a longer period of time. After this, the data have been analyzed. On the basis of the accomplished measurements and analyses, a series of mental models of web users for global (educational) purposes have been developed (e.g., the metamodel of thought hierarchy of web users, the segmentation model of web users), and the users were profiled in four different groups (adventurers, observers, applicable, and know-alls). The article concludes with the construction of a new knowledge management solution called multidimensional rank thesaurus.

The cost‐effectiveness analysis of information retrieval and dissemination systems

AbstractA distinction is made between cost‐effectiveness analysis and cost‐benefits analysis as applied to information systems; and the relationship between costs, performance, and benefits is discussed. Some factors influencing the cost‐effectiveness of retrieval and dissemination systems are identified. Various aspects of system operation that are susceptible to cost‐effectiveness analysis are discussed, including system coverage, indexing policies and procedures, system vocabulary, searching procedures, and mode of interaction between system and user. Possible tradeoffs between input and output costs, and the effects of these tradeoffs on cost‐effectiveness are presented.