Natural Language Processing Approaches Research Articles

Sector-level sentiment analysis with deep learning

This paper presents new machine learning methods in the context of natural language processing (NLP) to extract useful information from financial news. Traditional NLP approaches, which are based on the use of lexicons or standard machine learning algorithms, ignore the importance of word position and combinations in texts, thereby resulting in low performance. More recently, NLP empowered by deep learning has achieved remarkable results in various tasks such as sentiment analysis. This paper proposes a deep learning solution for sentiment analysis, which is trained exclusively on financial news and combines multiple recurrent neural networks. Subsequently, our sentiment analysis models are enhanced using a semi-supervised learning method that relies on the detection and correction of presumably mislabeled data. The performance of our proposed solution compared favorably against both traditional and state-of-the-art models based on its performance on previously unseen tweet data. Additionally, this study provides a novel research on the prediction of specific economic sectors affected by news articles. Finally, we propose an ensemble of sentiment and sector models to provide a sector-level sentiment analysis with potential applications in the context of sector fund indices.

From Text Representation to Financial Market Prediction: A Literature Review

News dissemination in social media causes fluctuations in financial markets. (Scope) Recent advanced methods in deep learning-based natural language processing have shown promising results in financial market analysis. However, understanding how to leverage large amounts of textual data alongside financial market information is important for the investors’ behavior analysis. In this study, we review over 150 publications in the field of behavioral finance that jointly investigated natural language processing (NLP) approaches and a market data analysis for financial decision support. This work differs from other reviews by focusing on applied publications in computer science and artificial intelligence that contributed to a heterogeneous information fusion for the investors’ behavior analysis. (Goal) We study various text representation methods, sentiment analysis, and information retrieval methods from heterogeneous data sources. (Findings) We present current and future research directions in text mining and deep learning for correlation analysis, forecasting, and recommendation systems in financial markets, such as stocks, cryptocurrencies, and Forex (Foreign Exchange Market).

An Assessment of Mentions of Adverse Drug Events on Social Media With Natural Language Processing: Model Development and Analysis.

BackgroundAdverse reactions to drugs attract significant concern in both clinical practice and public health monitoring. Multiple measures have been put into place to increase postmarketing surveillance of the adverse effects of drugs and to improve drug safety. These measures include implementing spontaneous reporting systems and developing automated natural language processing systems based on data from electronic health records and social media to collect evidence of adverse drug events that can be further investigated as possible adverse reactions.ObjectiveWhile using social media for collecting evidence of adverse drug events has potential, it is not clear whether social media are a reliable source for this information. Our work aims to (1) develop natural language processing approaches to identify adverse drug events on social media and (2) assess the reliability of social media data to identify adverse drug events.MethodsWe propose a collocated long short-term memory network model with attentive pooling and aggregated, contextual representation generated by a pretrained model. We applied this model on large-scale Twitter data to identify adverse drug event–related tweets. We conducted a qualitative content analysis of these tweets to validate the reliability of social media data as a means to collect such information.ResultsThe model outperformed a variant without contextual representation during both the validation and evaluation phases. Through the content analysis of adverse drug event tweets, we observed that adverse drug event–related discussions had 7 themes. Mental health–related, sleep-related, and pain-related adverse drug event discussions were most frequent. We also contrast known adverse drug reactions to those mentioned in tweets.ConclusionsWe observed a distinct improvement in the model when it used contextual information. However, our results reveal weak generalizability of the current systems to unseen data. Additional research is needed to fully utilize social media data and improve the robustness and reliability of natural language processing systems. The content analysis, on the other hand, showed that Twitter covered a sufficiently wide range of adverse drug events, as well as known adverse reactions, for the drugs mentioned in tweets. Our work demonstrates that social media can be a reliable data source for collecting adverse drug event mentions.

Mental Health Diagnoses are Not Associated With Indicators of Lower Quality Pain Care in Electronic Health Records of a National Sample of Veterans Treated in Veterans Health Administration Primary Care Settings

Prior research has demonstrated disparities in general medical care for patients with mental health conditions, but little is known about disparities in pain care. The objective of this retrospective cohort study was to determine whether mental health conditions are associated with indicators of pain care quality (PCQ) as documented by primary care clinicians in the Veterans Health Administration (VHA). We used natural language processing to analyze electronic health record data from a national sample of Veterans with moderate to severe musculoskeletal pain during primary care visits in the Fiscal Year 2017. Twelve PCQ indicators were annotated from clinician progress notes as present or absent; PCQ score was defined as the sum of these indicators. Generalized estimating equation Poisson models examined associations among mental health diagnosis categories and PCQ scores. The overall mean PCQ score across 135,408 person-visits was 8.4 (SD = 2.3). In the final adjusted model, post-traumatic stress disorder was associated with higher PCQ scores (RR = 1.006, 95%CI 1.002–1.010, P = .007). Depression, alcohol use disorder, other substance use disorder, schizophrenia, and bipolar disorder diagnoses were not associated with PCQ scores. Overall, results suggest that in this patient population, presence of a mental health condition is not associated with lower quality pain care. PerspectiveThis study used a natural language processing approach to analyze medical records to determine whether mental health conditions are associated with indicators of pain care quality as documented by primary care clinicians. Findings suggest that presence of a diagnosed mental health condition is not associated with lower quality pain care.

Training and intrinsic evaluation of lightweight word embeddings for the clinical domain in Spanish.

Resources for Natural Language Processing (NLP) are less numerous for languages different from English. In the clinical domain, where these resources are vital for obtaining new knowledge about human health and diseases, creating new resources for the Spanish language is imperative. One of the most common approaches in NLP is word embeddings, which are dense vector representations of a word, considering the word's context. This vector representation is usually the first step in various NLP tasks, such as text classification or information extraction. Therefore, in order to enrich Spanish language NLP tools, we built a Spanish clinical corpus from waiting list diagnostic suspicions, a biomedical corpus from medical journals, and term sequences sampled from the Unified Medical Language System (UMLS). These three corpora can be used to compute word embeddings models from scratch using Word2vec and fastText algorithms. Furthermore, to validate the quality of the calculated embeddings, we adapted several evaluation datasets in English, including some tests that have not been used in Spanish to the best of our knowledge. These translations were validated by two bilingual clinicians following an ad hoc validation standard for the translation. Even though contextualized word embeddings nowadays receive enormous attention, their calculation and deployment require specialized hardware and giant training corpora. Our static embeddings can be used in clinical applications with limited computational resources. The validation of the intrinsic test we present here can help groups working on static and contextualized word embeddings. We are releasing the training corpus and the embeddings within this publication1.

A Natural Language Processing (NLP) Evaluation on COVID-19 Rumour Dataset Using Deep Learning Techniques.

Context and Background: Since December 2019, the coronavirus (COVID-19) epidemic has sparked considerable alarm among the general community and significantly affected societal attitudes and perceptions. Apart from the disease itself, many people suffer from anxiety and depression due to the disease and the present threat of an outbreak. Due to the fast propagation of the virus and misleading/fake information, the issues of public discourse alter, resulting in significant confusion in certain places. Rumours are unproven facts or stories that propagate and promote sentiments of prejudice, hatred, and fear. Objective. The study's objective is to propose a novel solution to detect fake news using state-of-the-art machines and deep learning models. Furthermore, to analyse which models outperformed in detecting the fake news. Method. In the research study, we adapted a COVID-19 rumours dataset, which incorporates rumours from news websites and tweets, together with information about the rumours. It is important to analyse data utilizing Natural Language Processing (NLP) and Deep Learning (DL) approaches. Based on the accuracy, precision, recall, and the f1 score, we can assess the effectiveness of the ML and DL algorithms. Results. The data adopted from the source (mentioned in the paper) have collected 9200 comments from Google and 34,779 Twitter postings filtered for phrases connected with COVID-19-related fake news. Experiment 1. The dataset was assessed using the following three criteria: veracity, stance, and sentiment. In these terms, we have different labels, and we have applied the DL algorithms separately to each term. We have used different models in the experiment such as (i) LSTM and (ii) Temporal Convolution Networks (TCN). The TCN model has more performance on each measurement parameter in the evaluated results. So, we have used the TCN model for the practical implication for better findings. Experiment 2. In the second experiment, we have used different state-of-the-art deep learning models and algorithms such as (i) Simple RNN; (ii) LSTM + Word Embedding; (iii) Bidirectional + Word Embedding; (iv) LSTM + CNN-1D; and (v) BERT. Furthermore, we have evaluated the performance of these models on all three datasets, e.g., veracity, stance, and sentiment. Based on our second experimental evaluation, the BERT has a superior performance over the other models compared.

Emotions and Incivility in Vaccine Mandate Discourse: Natural Language Processing Insights.

BackgroundDespite vaccine availability, vaccine hesitancy has inhibited public health officials’ efforts to mitigate the COVID-19 pandemic in the United States. Although some US elected officials have responded by issuing vaccine mandates, others have amplified vaccine hesitancy by broadcasting messages that minimize vaccine efficacy. The politically polarized nature of COVID-19 information on social media has given rise to incivility, wherein health attitudes often hinge more on political ideology than science.ObjectiveTo the best of our knowledge, incivility has not been studied in the context of discourse regarding COVID-19 vaccines and mandates. Specifically, there is little focus on the psychological processes that elicit uncivil vaccine discourse and behaviors. Thus, we investigated 3 psychological processes theorized to predict discourse incivility—namely, anxiety, anger, and sadness.MethodsWe used 2 different natural language processing approaches: (1) the Linguistic Inquiry and Word Count computational tool and (2) the Google Perspective application programming interface (API) to analyze a data set of 8014 tweets containing terms related to COVID-19 vaccine mandates from September 14, 2021, to October 1, 2021. To collect the tweets, we used the Twitter API Tweet Downloader Tool (version 2). Subsequently, we filtered through a data set of 375,000 vaccine-related tweets using keywords to extract tweets explicitly focused on vaccine mandates. We relied on the Linguistic Inquiry and Word Count computational tool to measure the valence of linguistic anger, sadness, and anxiety in the tweets. To measure dimensions of post incivility, we used the Google Perspective API.ResultsThis study resolved discrepant operationalizations of incivility by introducing incivility as a multifaceted construct and explored the distinct emotional processes underlying 5 dimensions of discourse incivility. The findings revealed that 3 types of emotions—anxiety, anger, and sadness—were uniquely associated with dimensions of incivility (eg, toxicity, severe toxicity, insult, profanity, threat, and identity attacks). Specifically, the results showed that anger was significantly positively associated with all dimensions of incivility (all P<.001), whereas sadness was significantly positively related to threat (P=.04). Conversely, anxiety was significantly negatively associated with identity attack (P=.03) and profanity (P=.02).ConclusionsThe results suggest that our multidimensional approach to incivility is a promising alternative to understanding and intervening in the psychological processes underlying uncivil vaccine discourse. Understanding specific emotions that can increase or decrease incivility such as anxiety, anger, and sadness can enable researchers and public health professionals to develop effective interventions against uncivil vaccine discourse. Given the need for real-time monitoring and automated responses to the spread of health information and misinformation on the web, social media platforms can harness the Google Perspective API to offer users immediate, automated feedback when it detects that a comment is uncivil.

Text-based automatic personality prediction: a bibliographic review

Personality detection is an old topic in psychology and Automatic Personality Prediction (or Perception) (APP) is the automated (computationally) forecasting of the personality on different types of human generated/exchanged contents (such as text, speech, image, video). The principal objective of this study is to offer a shallow (overall) review of natural language processing approaches on APP since 2010. With the advent of deep learning and following it transfer-learning and pre-trained model in NLP, APP research area has been a hot topic, so in this review, methods are categorized into three; pre-trained independent, pre-trained model based, multimodal approaches. Also, to achieve a comprehensive comparison, reported results are informed by datasets.

Interpreting low-carbon transition at the subnational level: Evidence from China using a Natural Language Processing approach

Subnational governments play an important role in low-carbon transitions around the world. However, the concept of low-carbon transition requires detailed interpretations. Our study sheds light on this important gap by drawing evidence from China's low-carbon provincial and city pilots (LCPCs) using three natural language processing algorithms. Our study found that low-carbon transition was interpreted collectively as an economy-centered process. Five primary foci and the corresponding clusters were identified, which indicate the preferred transition pathways of the LCPCs and imply that the pilots recognized no “one size fits all” transitions. Innovative interpretations were found among LCPCs, which were largely embedded in local contexts and tied to unique local resources and actors. However, innovative interpretations with local embeddedness might not translate into effective policies. And "just transition" has been largely missing from the interpretation. The concept of low-carbon transition may still remain vague to the subnational policymakers at a more detailed level.

PhisherCop: Developing an NLP-Based Automated Tool for Phishing Detection

Phishing poses a major security risk to organizations and individuals leading to loss of billions of dollars yearly. While risk communication serves as a tool to mitigate phishing attempts, it is imperative to create automated phishing detection tools. Numerous Natural Language Processing (NLP) approaches have been deployed to tackle phishing. However, phishing attempts continue to increase exponentially, which reiterates the need for more effective approach. To address this, we have developed an anti-phishing tool called; PhisherCop. PhisherCop is built upon Stochastic Gradient Descent classifier (SGD) and Support Vector Classifier (SVC) which showed an average accuracy of 96%, performing better than six other popular classifiers including: Decision Tree, Logistics Regression, Random Forest, Gradient Boosting Classifier, K-Nearest Neighbors and Multinomial Naive Bayes. Our tool is significant in distinguishing between phishing and legitimate content both over emails and text messages.

Obfuscation detection in Android applications using deep learning

Malware is often hidden in illegitimately cloned software. Android, with over two billions active devices, is one of the most affected platforms because code cloning is quite simple and there are several not controlled markets. Obfuscation is both a cause and a solution to this scenario: a cause because obfuscated malware is harder to detect, a solution because obfuscation of legitimate applications makes code cloning more difficult. A deeper understanding of the obfuscation techniques would lead to more effective and aware use. In the literature, there are few methods of obfuscation detection with limited accuracy. Manual reverse engineering is too time-consuming to achieve this purpose, we need faster and automated techniques.In this work, we propose several deep learning models that can detect and classify the presence of obfuscation in Android applications. In addition to classical ML methods, we leverage natural language processing or image recognition approaches, then with a hybrid model, we exploit the best of each approach. Tests over a large dataset, made using different obfuscation tools, showed improvements compared to previous obfuscation detection methods. We target four obfuscation classes: identifier renaming, string encryption, reflection and class encryption, achieving an average F-measure of 0.985.

Natural Language Processing in Nephrology.

Unstructured data in the electronic health records contain essential patient information. Natural language processing (NLP), teaching a computer to read, allows us to tap into these data without needing the time and effort of manual chart abstraction. The core first step for all NLP algorithms is preprocessing the text to identify the core words that differentiate the text while filtering out the noise. Traditional NLP uses a rule-based approach, applying grammatical rules to infer meaning from the text. Newer NLP approaches use machine learning/deep learning which can infer meaning without explicitly being programmed. NLP use in nephrology research has focused on identifying distinct disease processes, such as CKD, and extraction of patient-oriented outcomes such as symptoms with high sensitivity. NLP can identify patient features from clinical text associated with acute kidney injury and progression of CKD. Lastly, inclusion of features extracted using NLP improved the performance of risk-prediction models compared to models that only use structured data. Implementation of NLP algorithms has been slow, partially hindered by the lack of external validation of NLP algorithms. However, NLP allows for extraction of key patient characteristics from free text, an infrequently used resource in nephrology.

Using textual bug reports to predict the fault category of software bugs

Debugging is a time-consuming and expensive process. Developers have to select appropriate tools, methods and approaches in order to efficiently reproduce, localize and fix bugs. These choices are based on the developers’ assessment of the type of fault for a given bug report. This paper proposes a machine learning (ML) based approach that predicts the fault type for a given textual bug report. We built a dataset from 70+ projects for training and evaluation of our approach. Further, we performed a user study to establish a baseline for non-expert human performance on this task. Our models, incorporating our custom preprocessing approaches, reach up to 0.69% macro average F1 score on this bug classification problem. We demonstrate inter-project transferability of our approach. Further, we identify and discuss issues and limitations of ML classification approaches applied on textual bug reports. Our models can support researchers in data collection efforts, as for example bug benchmark creation. In future, such models could aid inexperienced developers in debugging tool selection, helping save time and resources. • Natural language processing approach predicting the fault type from bug reports. • A data set of 762 labeled bug reports from 75 GitHub Projects. • A user survey to establish a performance baseline for fault type classification. • Machine learning and ensemble methods classifying bug reports.

TFIDF-Random Forest: Prediction of Aptamer-Protein Interacting Pairs.

Aptamers are short, single-stranded oligonucleotides or peptides generated from in vitro selection to selectively bind with various molecules. Due to their molecular recognition capability for proteins, aptamers are becoming promising reagents in new drug development. Aptamers can fold into specific spatial configuration that bind to certain targets with extremely high specificity. The ability of aptamers to reversibly bind proteins has generated increasing interest in using them to facilitate controlled release of therapeutic biomolecules. In-vitro selection experiments to produce the aptamer-protein binding pairs is very complex and MD/MM in-silico experiments can be computationally expensive. In this study, we introduce a natural language processing approach for data-driven computational selection. We compared our method to the sequential model with the embedding layer, applied in the literature. We transformed the DNA/RNA and protein sequences into text format using a sliding window approach. This methodology showed that efficiency was notably higher than those observed from the literature. This indicates that our preliminary model has marked improvement over previous models which brings us closer to a data-driven computational selection method.

Natural Language Processing Approaches for Automated Multilevel and Multiclass Classification of Breast Lesions on Free-Text Cytopathology Reports.

The extensive growth and use of electronic health records (EHRs) and extending medical literature have led to huge opportunities to automate the extraction of relevant clinical information that helps in concise and effective clinical decision support. However, processing such information has traditionally been dependent on labor-intensive processes with human errors such as fatigue, oversight, and interobserver variability. Hence, this study aims at the processing of EHRs and performing multilevel and multiclass classification by fetching dominant characteristic features that are sufficient to detect and differentiate various types of breast lesions. In this study, unstructured EHRs on breast lesions obtained through fine-needle aspiration cytology technique are considered. The raw text was normalized into structured tabular form and converted to scores by performing sentiment analysis that helps to decide the total polarity or class label of the EHR. Supervised machine learning approaches, namely random forest and feed-forward neural network trained using Levenberg-Marquardt training function, are used for classification of the collected EHR data set containing 2,879 records that are split in the ratio of 80:20 as training and testing data sets, respectively. Random forest and feed-forward neural network classifiers gave the best performance with an accuracy of 99.36%, an overall receiver operating characteristic-area under the curve of 99.2%, a correlation with ground truth of 98.3%, and a histopathologic correlation of 98.6%. Natural language processing has huge potential to automate the extraction of clinical features from breast lesions. The proposed multilevel and multiclass classification approach is used to classify 13 different types of breast lesions with 20 different labels into five classes to decide the type of treatment that should be given to patients by a physician or oncologist.

Housing safety and health academic and public opinion mining from 1945 to 2021: PRISMA, cluster analysis, and natural language processing approaches.

Housing safety and health problems threaten owners' and occupiers' safety and health. Nevertheless, there is no systematic review on this topic to the best of our knowledge. This study compared the academic and public opinions on housing safety and health and reviewed 982 research articles and 3,173 author works on housing safety and health published in the Web of Science Core Collection. PRISMA was used to filter the data, and natural language processing (NLP) was used to analyze emotions of the abstracts. Only 16 housing safety and health articles existed worldwide before 1998 but increased afterward. U.S. scholars published most research articles (30.76%). All top 10 most productive countries were developed countries, except China, which ranked second (16.01%). Only 25.9% of institutions have inter-institutional cooperation, and collaborators from the same institution produce most work. This study found that most abstracts were positive (n = 521), but abstracts with negative emotions attracted more citations. Despite many industries moving toward AI, housing safety and health research are exceptions as per articles published and Tweets. On the other hand, this study reviewed 8,257 Tweets to compare the focus of the public to academia. There were substantially more housing/residential safety (n = 8198) Tweets than housing health Tweets (n = 59), which is the opposite of academic research. Most Tweets about housing/residential safety were from the United Kingdom or Canada, while housing health hazards were from India. The main concern about housing safety per Twitter includes finance, people, and threats to housing safety. By contrast, people mainly concerned about costs of housing health issues, COVID, and air quality. In addition, most housing safety Tweets were neutral but positive dominated residential safety and health Tweets.

Detecting non-natural language artifacts for de-noising bug reports

Textual documents produced in the software engineering process are a popular target for natural language processing (NLP) and information retrieval (IR) approaches. However, issue tickets often contain artifacts such as code snippets, log outputs and stack traces. These artifacts not only inflate the issue ticket sizes, but also can this noise constitute a real problem for some NLP approaches, and therefore has to be removed in the pre-processing of some approaches. In this paper, we present a machine learning based approach to classify textual content into natural language and non-natural language artifacts at line level. We show how data from GitHub issue trackers can be used for automated training set generation, and present a custom preprocessing approach for the task of artifact removal. The training sets are automatically created from Markdown annotated issue tickets and project documentation files. We use these generated training sets to train a Markdown agnostic model that is able to classify un-annotated content. We evaluate our approach on issue tickets from projects written in C++, Java, JavaScript, PHP, and Python. Our approach achieves ROC-AUC scores between 0.92 and 0.96 for language-specific models. A multi-language model trained on the issue tickets of all languages achieves ROC-AUC scores between 0.92 and 0.95. The provided models are intended to be used as noise reduction pre-processing steps for NLP and IR approaches working on issue tickets.

Identifying Patients Who Meet Criteria for Genetic Testing of Hereditary Cancers Based on Structured and Unstructured Family Health History Data in the Electronic Health Record: Natural Language Processing Approach.

BackgroundFamily health history has been recognized as an essential factor for cancer risk assessment and is an integral part of many cancer screening guidelines, including genetic testing for personalized clinical management strategies. However, manually identifying eligible candidates for genetic testing is labor intensive.ObjectiveThe aim of this study was to develop a natural language processing (NLP) pipeline and assess its contribution to identifying patients who meet genetic testing criteria for hereditary cancers based on family health history data in the electronic health record (EHR). We compared an algorithm that uses structured data alone with structured data augmented using NLP.MethodsAlgorithms were developed based on the National Comprehensive Cancer Network (NCCN) guidelines for genetic testing for hereditary breast or ovarian and colorectal cancers. The NLP-augmented algorithm uses both structured family health history data and the associated unstructured free-text comments. The algorithms were compared with a reference standard of 100 patients with a family health history in the EHR.ResultsRegarding identifying the reference standard patients meeting the NCCN criteria, the NLP-augmented algorithm compared with the structured data algorithm yielded a significantly higher recall of 0.95 (95% CI 0.9-0.99) versus 0.29 (95% CI 0.19-0.40) and a precision of 0.99 (95% CI 0.96-1.00) versus 0.81 (95% CI 0.65-0.95). On the whole data set, the NLP-augmented algorithm extracted 33.6% more entities, resulting in 53.8% more patients meeting the NCCN criteria.ConclusionsCompared with the structured data algorithm, the NLP-augmented algorithm based on both structured and unstructured family health history data in the EHR increased the number of patients identified as meeting the NCCN criteria for genetic testing for hereditary breast or ovarian and colorectal cancers.

Intelligent RFQ Summarization Using Natural Language Processing, Text Mining, and Machine Learning Techniques

Request for quotation (RFQ) is a lengthy document soliciting vendor products and services according to rigid specifications. This research develops an integrated natural language processing (NLP), text mining, and machine learning approach for intelligent RFQ summarization. Over 1,300 power transformer RFQ requests are used to build a word-embedding model for training and testing. Domain keywords are extracted using N-gram TF-IDF. The method automatically extracts essential specifications such as voltage, capacity, and impedance from RFQs using text analytics. The K-means algorithm groups the sentences of each specification. The TextRank algorithm identifies important sentences of all specifications to generate RFQ summaries. The summarization system helps engineers shorten the time to identify all specifications and reduces the risk of missing important requirements during manual RFQ reading. The system helps improve the complex product design for manufacturers and improve the cost estimation and competitiveness of quotations in a highly competitive marketplace.

Scholarly knowledge graphs through structuring scholarly communication: a review

The necessity for scholarly knowledge mining and management has grown significantly as academic literature and its linkages to authors produce enormously. Information extraction, ontology matching, and accessing academic components with relations have become more critical than ever. Therefore, with the advancement of scientific literature, scholarly knowledge graphs have become critical to various applications where semantics can impart meanings to concepts. The objective of study is to report a literature review regarding knowledge graph construction, refinement and utilization in scholarly domain. Based on scholarly literature, the study presents a complete assessment of current state-of-the-art techniques. We presented an analytical methodology to investigate the existing status of scholarly knowledge graphs (SKG) by structuring scholarly communication. This review paper investigates the field of applying machine learning, rule-based learning, and natural language processing tools and approaches to construct SKG. It further presents the review of knowledge graph utilization and refinement to provide a view of current research efforts. In addition, we offer existing applications and challenges across the board in construction, refinement and utilization collectively. This research will help to identify frontier trends of SKG which will motivate future researchers to carry forward their work.