RESUMEN
This article proposes a novel lexicon-based unsupervised sentiment analysis method to measure the "hope" and "fear" for the 2022 Ukrainian-Russian Conflict. Reddit.com is utilized as the main source of human reactions to daily events during nearly the first 3 months of the conflict. The top 50 "hot" posts of six different subreddits about Ukraine and news (Ukraine, worldnews, Ukraina, UkrainianConflict, UkraineWarVideoReport, and UkraineWarReports) along with their relative comments are scraped every day between 10th of May and 28th of July, and a novel data set is created. On this corpus, multiple analyzes, such as (1) public interest, (2) Hope/Fear score, and (3) stock price interaction, are employed. We use a dictionary approach, which scores the hopefulness of every submitted user post. The Latent Dirichlet Allocation (LDA) algorithm of topic modeling is also utilized to understand the main issues raised by users and what are the key talking points. Experimental analysis shows that the hope strongly decreases after the symbolic and strategic losses of Azovstal (Mariupol) and Severodonetsk. Spikes in hope/fear, both positives and negatives, are present not only after important battles, but also after some non-military events, such as Eurovision and football games.
RESUMEN
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.
RESUMEN
Text mining (TM) in the field of biology is fast becoming a routine analysis for the extraction and curation of biological entities (e.g., genes, proteins, simple chemicals) as well as their relationships. Due to the wide applicability of TM in situations involving complex relationships, it is valuable to apply TM to the extraction of metabolic interactions (i.e., enzyme and metabolite interactions) through metabolic events. Here we present an integrated TM framework containing two modules for the extraction of metabolic events (Metabolic Event Extraction module-MEE) and for the construction of a metabolic interaction network (Metabolic Interaction Network Reconstruction module-MINR). The proposed integrated TM framework performed well based on standard measures of recall, precision and F-score. Evaluation of the MEE module using the constructed Metabolic Entities (ME) corpus yielded F-scores of 59.15% and 48.59% for the detection of metabolic events for production and consumption, respectively. As for the testing of the entity tagger for Gene and Protein (GP) and metabolite with the test corpus, the obtained F-score was greater than 80% for the Superpathway of leucine, valine, and isoleucine biosynthesis. Mapping of enzyme and metabolite interactions through network reconstruction showed a fair performance for the MINR module on the test corpus with F-score >70%. Finally, an application of our integrated TM framework on a big-scale data (i.e., EcoCyc extraction data) for reconstructing a metabolic interaction network showed reasonable precisions at 69.93%, 70.63% and 46.71% for enzyme, metabolite and enzyme-metabolite interaction, respectively. This study presents the first open-source integrated TM framework for reconstructing a metabolic interaction network. This framework can be a powerful tool that helps biologists to extract metabolic events for further reconstruction of a metabolic interaction network. The ME corpus, test corpus, source code, and virtual machine image with pre-configured software are available at www.sbi.kmutt.ac.th/ preecha/metrecon.