Term Frequency-inverse Document Frequency Scores Research Articles

Hate Speech Detection Using NLP and Machine Learning

Abstract: The use of social media has been growing in an eccentric fashion making it a medium for sharing opinions, ideas, and thoughts of an individual with others. This has made things complex with what is considered a genuine comment or rather a hypocritic deliberative nuance to damage or incite hatred on an individual or a group belonging to a community, race, gender, nationality, etc, In this paper, the detection of hate speech with the use of sentiment polarity scores and the Term Frequency Inverse Document Frequency(TFIDF) scores with machine learning algorithms is to decrease the true negatives and false positives by the use of Natural Language Processing. The Machine Learning algorithms used are Logistic Regression and Random Forest Classifier. The phases of NLP are done to preprocess the tweets that are available on the Kaggle with about 25 thousand tweets from the social media giant “Twitter”. The processed tweets are then with the use of two ML Algorithms trained for vaderSentiment polarity scores and TFIDF scores from which metrics are obtained. The results of sentiment polarity scores(7 points) are less accurate in the detection of hate speech as compared to TFIDF scores(8 points).

How Does ChatGPT Use Source Information Compared With Google? A Text Network Analysis of Online Health Information.

How Does ChatGPT Use Source Information Compared With Google? A Text Network Analysis of Online Health Information.

Webpage Metadata Extration Using Machine Learning Techiniques

Abstract: This Python script defines a Flask web application enabling users to input a URL. The application fetches the webpage content and utilizes TF-IDF (Term Frequency-Inverse Document Frequency) analysis to extract information like the title, description, and top keywords. The / route renders an HTML template (index.html) for user input, while the /extract route handles a POST request, fetching the webpage content, extracting relevant information using TF-IDF analysis, and rendering the results in another HTML template (result.html). The TF-IDF process involves tokenizing the text, eliminating stopwords, and calculating TF-IDF scores for each term. The top 10 keywords are then extracted based on their TF-IDF scores. The script also incorporates error handling for cases where the webpage cannot be fetched or an exception occurs during the process.

Research Constituents and Trends in Smart Farming: An Analytical Retrospection from the Lens of Text Mining

Agriculture research began with the idea that local systems are interconnected. Thus, it was crucial to consider farmers, crops, and livestock. Smart farming arose with the Internet of Things (IoT) as people progressively digitized farming with new information technology. Academic and scientific groups innovate and commercialize IoT‐based agricultural products and solutions. Many public and private organizations also explore farming advancements. Therefore, we must stimulate communication and cooperation among the many farming industry actors to build smart agricultural standards and improve system and technology interoperability. The study analyzed 3,229 published articles from smart farming systems in the Scopus database between 2008 and 2022. The collected corpus is preprocessed through various steps, creating a bag of words. Text mining, latent semantic analysis, and network analysis are applied to the collected corpus to provide current research areas based on the key terms. The key terms are taken based on the term frequency‐inverse document frequency score. The research was experimented with using KNIME and VOSviewer. Finally, 10 current research areas are provided using K‐means clustering for a future researcher for deep insight. This study offered years of analysis, top journals, top authors, and leading countries contributing to smart farming. Research into the myriad issues facing smart farming could begin with these trends. This research helps future researchers understand smart farming and areas that need more attention. It also draws attention to research directions that could use further study.

RoBERTa-Assisted Outcome Prediction in Ovarian Cancer Cytoreductive Surgery Using Operative Notes.

Contemporary efforts to predict surgical outcomes focus on the associations between traditional discrete surgical risk factors. We aimed to determine whether natural language processing (NLP) of unstructured operative notes improves the prediction of residual disease in women with advanced epithelial ovarian cancer (EOC) following cytoreductive surgery. Electronic Health Records were queried to identify women with advanced EOC including their operative notes. The Term Frequency - Inverse Document Frequency (TF-IDF) score was used to quantify the discrimination capacity of sequences of words (n-grams) regarding the existence of residual disease. We employed the state-of-the-art RoBERTa-based classifier to process unstructured surgical notes. Discrimination was measured using standard performance metrics. An XGBoost model was then trained on the same dataset using both discrete and engineered clinical features along with the probabilities outputted by the RoBERTa classifier. The cohort consisted of 555 cases of EOC cytoreduction performed by eight surgeons between January 2014 and December 2019. Discrete word clouds weighted by n-gram TF-IDF score difference between R0 and non-R0 resection were identified. The words 'adherent' and 'miliary disease' best discriminated between the two groups. The RoBERTa model reached high evaluation metrics (AUROC .86; AUPRC .87, precision, recall, and F1 score of .77 and accuracy of .81). Equally, it outperformed models that used discrete clinical and engineered features and outplayed the performance of other state-of-the-art NLP tools. When the probabilities from the RoBERTa classifier were combined with commonly used predictors in the XGBoost model, a marginal improvement in the overall model's performance was observed (AUROC and AUPRC of .91, with all other metrics the same). We applied a sui generis approach to extract information from the abundant textual surgical data and demonstrated how it can be effectively used for classification prediction, outperforming models relying on conventional structured data. State-of-art NLP applications in biomedical texts can improve modern EOC care.

Natural Language Processing and Cognitive Networks Identify UK Insurers’ Trends in Investor Day Transcripts

The ability to spot key ideas, trends, and relationships between them in documents is key to financial services, such as banks and insurers. Identifying patterns across vast amounts of domain-specific reports is crucial for devising efficient and targeted supervisory plans, subsequently allocating limited resources where most needed. Today, insurance supervisory planning primarily relies on quantitative metrics based on numerical data (e.g., solvency financial returns). The purpose of this work is to assess whether Natural Language Processing (NLP) and cognitive networks can highlight events and relationships of relevance for regulators that supervise the insurance market, replacing human coding of information with automatic text analysis. To this aim, this work introduces a dataset of NIDT=829 investor transcripts from Bloomberg and explores/tunes 3 NLP techniques: (1) keyword extraction enhanced by cognitive network analysis; (2) valence/sentiment analysis; and (3) topic modelling. Results highlight that keyword analysis, enriched by term frequency-inverse document frequency scores and semantic framing through cognitive networks, could detect events of relevance for the insurance system like cyber-attacks or the COVID-19 pandemic. Cognitive networks were found to highlight events that related to specific financial transitions: The semantic frame of “climate” grew in size by +538% between 2018 and 2020 and outlined an increased awareness that agents and insurers expressed towards climate change. A lexicon-based sentiment analysis achieved a Pearson’s correlation of ρ=0.16 (p<0.001,N=829) between sentiment levels and daily share prices. Although relatively weak, this finding indicates that insurance jargon is insightful to support risk supervision. Topic modelling is considered less amenable to support supervision, because of a lack of results’ stability and an intrinsic difficulty to interpret risk patterns. We discuss how these automatic methods could complement existing supervisory tools in supporting effective oversight of the insurance market.

COMBINING NATURAL LANGUAGE PROCESSING METHOD AND MENDELIAN RANDOMIZATION TO ESTABLISH CAUSAL EFFECT OF BLOOD PRESSURE ON OUTCOMES

Objective: Mendelian rioritisedn (MR) uses genetic variants as instruments to estimate a causal effect of an exposure on an outcome. Specific conditions need to be satisfied to ensure the validity of the genetic variant as an instrumental variable (IV). Pleiotropy is one of the reasons for invalid IV and erroneous causal estimation. We applied natural language processing (NLP) on multidimensional GWAS datasets enriched with gene-expression information to select appropriate IVs for MR analysis. The selected set of IVs was then used in MR analysis to investigate the causal effect of blood pressure (BP) on cardiovascular outcomes (ischaemic stroke, atrial fibrillation (AF), coronary artery disease (CAD), and chronic kidney disease (CKD)). Design and method: We curated corpus of data by integrating GWAS results, gene expression, single nucleotide polymorphism (SNP) proxies and proximate genes on 2955 phenotypes. We used chromosome cytobands to partition the genome into documents, in which SNPs, genes, tissues and phenotype domains are treated as the terms. We subsequently calculated term frequency – inverse document frequency (TF-IDF) for all the terms in the corpus and identified the cytobands with the highest TF-IDF for BP. Independent non-pleiotropic SNPs associated with BP in the rioritised cytogenetic bands were selected as IVs for two-sample MR analysis. We performed inverse-variance weighted (IVW) MR and MR Egger analysis using GWAS meta-analysis systolic BP (SBP) estimates and the following outcomes: ischaemic stroke (N = 440328), AF (N = 1030836), CAD (N = 547261), CKD (N = 117165). Analyses were performed on the MR-Base platform. Results: We identified 10 non-pleiotropic, BP-associated SNPs from the top 10 cytobands with the highest TF-IDF scores for BP over the whole genome. Two-sample MR analyses showed a significant causal relationship between genetically determined greater SBP and higher risk of ischemic stroke, AF but not for CAD and CKD (ischemic stroke: IVW = 0.03 ± 0.01, P = 0.006; MR-Egger = 0.05 ± 0.02, P = 0.09; AF: IVW = 0.03 ± 0.006, P = 9.017E-07; MR-Egger = 0.04 ± 0.01, P = 0.01) (Figure). Conclusions: Our joint NLP and MR analyses allowed selection of a parsimonious set of instrumental variable SNPs for MR analyses and may be an option for high-throughput analyses across multiple traits.

Word-level text highlighting of medical texts for telehealth services

The medical domain is often subject to information overload. The digitization of healthcare, constant updates to online medical repositories, and increasing availability of biomedical datasets make it challenging to effectively analyze the data. This creates additional workload for medical professionals who are heavily dependent on medical data to complete their research and consult their patients. This paper aims to show how different text highlighting techniques can capture relevant medical context. This would reduce the doctors' cognitive load and response time to patients by facilitating them in making faster decisions, thus improving the overall quality of online medical services. Three different word-level text highlighting methodologies are implemented and evaluated. The first method uses Term Frequency - Inverse Document Frequency (TF-IDF) scores directly to highlight important parts of the text. The second method is a combination of TF-IDF scores, Word2Vec and the application of Local Interpretable Model-Agnostic Explanations to classification models. The third method uses neural networks directly to make predictions on whether or not a word should be highlighted. Our numerical study shows that the neural network approach is successful in highlighting medically-relevant terms and its performance is improved as the size of the input segment increases.

Natural language processing of head CT reports to identify intracranial mass effect: CTIME algorithm.

The Mortality Probability Model (MPM) is used in research and quality improvement to adjust for severity of illness and can also inform triage decisions. However, a limitation for its automated use or application is that it includes the variable "intracranial mass effect" (IME), which requires human engagement with the electronic health record (EHR). We developed and tested a natural language processing (NLP) algorithm to identify IME from CT head reports. We obtained initial CT head reports from adult patients who were admitted to the ICU from our ED between 10/2013 and 9/2016. Each head CT head report was labeled yes/no IME by at least two of five independent labelers. The reports were then randomly divided 80/20 into training and test sets. All reports were preprocessed to remove linguistic and style variability, and a dictionary was created to map similar common terms. We tested three vectorization strategies: Term Frequency-Inverse Document frequency (TF-IDF), Word2Vec, and Universal Sentence Encoder to convert the report text to a numerical vector. This vector served as the input to a classification-tree-based ensemble machine learning algorithm (XGBoost). After training, model performance was assessed in the test set using the area under the receiver operating characteristic curve (AUROC). We also divided the continuous range of scores into positive/inconclusive/negative categories for IME. Of the 1202 CT reports in the training set, 308 (25.6%) reports were manually labeled as "yes" for IME. Of the 355 reports in the test set, 108 (30.4%) were labeled as "yes" for IME. The TF-IDF vectorization strategy as an input for the XGBoost model had the best AUROC:-- 0.9625 (95% CI 0.9443-0.9807). TF-IDF score categories were defined and had the following likelihood ratios: "positive" (TF-IDF score>0.5) LR=24.59; "inconclusive" (TF-IDF 0.05-0.5) LR=0.99; and "negative" (TF-IDF<0.05) LR=0.05. 82% of reports were classified as either "positive" or "negative". In the test set, only 4 of 199 (2.0%) reports with a "negative" classification were false negatives and only 8 of 93 (8.6%) reports classified as "positive" were false positives. NLP can accurately identify IME from free-text reports of head CTs in approximately 80% of records, adequate to allow automatic calculation of MPM based on EHR data for many applications.

The Evaluation of Accuracy Performance in an Enhanced Embedded Feature Selection for Unstructured Text Classification

Text documents are unstructured and high dimensional. Effective feature selection is required to select the most important and significant feature from the sparse feature space. Thus, this paper proposed an embedded feature selection technique based on Term Frequency-Inverse Document Frequency (TF-IDF) and Support Vector Machine-Recursive Feature Elimination (SVM-RFE) for unstructured and high dimensional text classificationhis technique has the ability to measure the feature’s importance in a high-dimensional text document. In addition, it aims to increase the efficiency of the feature selection. Hence, obtaining a promising text classification accuracy. TF-IDF act as a filter approach which measures features importance of the text documents at the first stage. SVM-RFE utilized a backward feature elimination scheme to recursively remove insignificant features from the filtered feature subsets at the second stage. This research executes sets of experiments using a text document retrieved from a benchmark repository comprising a collection of Twitter posts. Pre-processing processes are applied to extract relevant features. After that, the pre-processed features are divided into training and testing datasets. Next, feature selection is implemented on the training dataset by calculating the TF-IDF score for each feature. SVM-RFE is applied for feature ranking as the next feature selection step. Only top-rank features will be selected for text classification using the SVM classifier. Based on the experiments, it shows that the proposed technique able to achieve 98% accuracy that outperformed other existing techniques. In conclusion, the proposed technique able to select the significant features in the unstructured and high dimensional text document.

Early detection of cognitive impairment by natural language among outpatients and community dwellers

Background: Although early detection of cognitive decline has a significant relation to improving the quality of life of dementia patients, this early detection has been difficult due to requires of neuropsychological tests which people generally take when they notice their cognitive impairment. The timing of patients’ notice was reported to be worse cognitive decline already, thus, we aimed to determine if cognitive impairment from a short interview by using Natural Language Processing approach. Methods: The present study used cross-sectional analysis among elderly outpatients and community-dwelling elderly from Septuagenarians, Octogenarians, Nonagenarians Investigation with Centenarians (SONIC) study. Cognitive decline was assessed by Telephone Interview of Cognitive Status for Japanese (TICS-J) and modeled as a binary outcome (cut-off <33 points). Natural language data was collected by semistructured interviews about health conditions and cognitive orientation in space, time, and place. We used an open-source text segmentation library to parse natural language text into bag-of-words and term frequency-inverse document frequency (TF-IDF) representations. Results: There were 38 (19.9%) outpatients and 153 (80.1%) community dwellers, and 60 (31.4%) participants were defined as cognitive impairment. The maximized TF-IDF score was 0.49 in cognitive orientation in time questions. In this question, participants without cognitive impairment could not calculate the score. There were no significant differences in TF-IDF scores between participants with and without cognitive impairment. Conclusions: Elderly without cognitive impairment might not have an episode about cognitive orientation in time, and this may help for early detection of cognitive impairment

CERC: an interactive content extraction, recognition, and construction tool for clinical and biomedical text

BackgroundAutomated summarization of scientific literature and patient records is essential for enhancing clinical decision-making and facilitating precision medicine. Most existing summarization methods are based on single indicators of relevance, offer limited capabilities for information visualization, and do not account for user specific interests. In this work, we develop an interactive content extraction, recognition, and construction system (CERC) that combines machine learning and visualization techniques with domain knowledge for highlighting and extracting salient information from clinical and biomedical text.MethodsA novel sentence-ranking framework multi indicator text summarization, MINTS, is developed for extractive summarization. MINTS uses random forests and multiple indicators of importance for relevance evaluation and ranking of sentences. Indicative summarization is performed using weighted term frequency-inverse document frequency scores of over-represented domain-specific terms. A controlled vocabulary dictionary generated using MeSH, SNOMED-CT, and PubTator is used for determining relevant terms. 35 full-text CRAFT articles were used as the training set. The performance of the MINTS algorithm is evaluated on a test set consisting of the remaining 32 full-text CRAFT articles and 30 clinical case reports using the ROUGE toolkit.ResultsThe random forests model classified sentences as “good” or “bad” with 87.5% accuracy on the test set. Summarization results from the MINTS algorithm achieved higher ROUGE-1, ROUGE-2, and ROUGE-SU4 scores when compared to methods based on single indicators such as term frequency distribution, position, eigenvector centrality (LexRank), and random selection, p < 0.01. The automatic language translator and the customizable information extraction and pre-processing pipeline for EHR demonstrate that CERC can readily be incorporated within clinical decision support systems to improve quality of care and assist in data-driven and evidence-based informed decision making for direct patient care.ConclusionsWe have developed a web-based summarization and visualization tool, CERC (https://newton.isye.gatech.edu/CERC1/), for extracting salient information from clinical and biomedical text. The system ranks sentences by relevance and includes features that can facilitate early detection of medical risks in a clinical setting. The interactive interface allows users to filter content and edit/save summaries. The evaluation results on two test corpuses show that the newly developed MINTS algorithm outperforms methods based on single characteristics of importance.

Following the Leader: An Analysis of Leadership and Conformity in Business Meetings

Background: Past research has established the importance of discursive leadership in professional communication, but it has not systematically examined how conformity behaviors emerge as a potentially undesirable consequence of discursive leadership. Literature review: Review of the literature on the centrality of communication in leadership processes and conformity behavior suggest a void of analytic tools to adequately examine the negative consequences of discursive leadership. Research question: Are later interlocutors more likely to speak similarly to earlier ones if the earlier interlocutors occupy a more central position in the conversation network? Methodology: Based on 32,000 words of a transcribed meeting corpus, we measured conformity behaviors using Term Frequency-Inverse Document Frequency scores, which are widely used in the information retrieval setting. We also operationalized the strength of discursive leadership as a positional centrality measure in the conversation network using a matrix algebra approach in social network analysis. Results: Findings support the hypothesis that discursive leadership is associated with conformity in language aligned toward discursive leaders' opinions. Conclusions: This study makes theoretical advances in understanding leadership construction and conformity behaviors between leaders and followers using empirical, authentic meeting data. We also give business people an applied understanding of the process of discursive leadership, which may help them to improve communication efficacy in their organizations by reducing overly conforming behaviors. We recommend that future research include more diverse participants and be combined with a survey to supplement the conversation data.

Enhanced Data Mining Technique to Measure Satisfaction Degree of Social Media Users of Xeljanz Drug

In the recent times, social media has become important in the field of health care as a major resource of valuable health information. Social media can provide massive amounts of data in real-time through user interaction, and this data can be analysed to reflect the harms and benefits of treatment by using the personal health experiences of users to improve health outcomes. In this study, we propose an enhanced data mining framework for analysing user opinions on Twitter and on a health-care forum. The proposed framework measures the degree of satisfaction of consumers regarding the drug Xeljanz, which is used to treat rheumatoid arthritis. The proposed framework is based on seven steps distributed in two phases. The first phase involves aggregating data related to the drug Xeljanz. This data is pre-processed to produce a list of words with a term frequency-inverse document frequency score. The word list is then classified into the following three categories: positive, negative and neutral. The second phase involves modelling social media posts using network analysis, identifying sub-graphs, calculating average opinions and detecting influential users. The results showed 77.3% user satisfaction with Xeljanz. Positive opinions were especially pronounced among users who switched to Xeljanz based on advice from a physician. Negative opinions of Xeljanz typically pertained to the high cost of the drug.

PhishSKaPe: A Content based Approach to Escape Phishing Attacks

Abstract A type of cyberattack, viz. phishing attack exposes a user’s credentials to malicious entities leading to breach of privacy. The proposed approach involves employing a search engine to match the domain name of the site under scrutiny with the sites that come up as result of our search query. Here, our search query is comprised of features viz. domain name and keywords consisting of title, meta description data and body text. We initially see how conventional TF-IDF (Term Frequency-Inverse Document Frequency) performs in classifying websites as phishing or legitimate. We then apply a weighted heuristic proposed in detail in this report by assigning different weights to different tag data and manipulate the TF-IDF score accordingly to improve the performance of our phishing detector. Experiments results illustrate that proposed search based method with weighted TF-IDF performs better than conventional TF-IDF.

Monitoring ‘Cyber Related’ Discussions in Online Social Platforms

As the use of social platforms continues to evolve, in areas such as cyber-security and defence, it has become imperative to develop adaptive methods for tracking, identifying and investigating cyber-related activities on these platforms. This paper introduces a new approach for detecting cyber-related discussions in online social platforms using a candidate set of terms that are representative of the cyber domain. The objective of this paper is to create a cyber lexicon with cyber-related terms that is applicable to the automatic detection of cyber activities across various online platforms. The method presented in this paper applies natural language processing techniques to representative data from multiple social platform types such as Reddit, Stack overflow, twitter and cyberwar news to extract candidate terms for a generic cyber lexicon. In selecting the candidate terms, we introduce the APMIS Aggregated Pointwise Mutual Information Score in comparison with the Term Frequency-Term Degree Ratio (FDR Score) and Term Frequency-Inverse Document Frequency Score (TF-IDF Score). These scoring mechanisms are robust to account for term frequency, term relevance and mutual dependence between terms. Finally, we evaluate the performance of the cyber lexicon by measuring its precision of in classifying discussions as 'Cyber-Related' or 'Non-Cyber-Related'.