Phrase Matching Research Articles

FLMatchQA: a recursive neural network-based question answering with customized federated learning model.

More sophisticated data access is possible with artificial intelligence (AI) techniques such as question answering (QA), but regulations and privacy concerns have limited their use. Federated learning (FL) deals with these problems, and QA is a viable substitute for AI. The utilization of hierarchical FL systems is examined in this research, along with an ideal method for developing client-specific adapters. The User Modified Hierarchical Federated Learning Model (UMHFLM) selects local models for users' tasks. The article suggests employing recurrent neural network (RNN) as a neural network (NN) technique for learning automatically and categorizing questions based on natural language into the appropriate templates. Together, local and global models are developed, with the worldwide model influencing local models, which are, in turn, combined for personalization. The method is applied in natural language processing pipelines for phrase matching employing template exact match, segmentation, and answer type detection. The (SQuAD-2.0), a DL-based QA method for acquiring knowledge of complicated SPARQL test questions and their accompanying SPARQL queries across the DBpedia dataset, was used to train and assess the model. The SQuAD2.0 datasets evaluate the model, which identifies 38 distinct templates. Considering the top two most likely templates, the RNN model achieves template classification accuracy of 92.8% and 61.8% on the SQuAD2.0 and QALD-7 datasets. A study on data scarcity among participants found that FL Match outperformed BERT significantly. A MAP margin of 2.60% exists between BERT and FL Match at a 100% data ratio and an MRR margin of 7.23% at a 20% data ratio.

Studying Mandarin Tone sandhi in interaction

The Mandarin low-dipping tone (T3) undergoes an alteration, tone sandhi, when followed by another T3. The resulting F0 is superficially the same as that of the rising tone (T2). We investigated how Mandarin speakers adapt their speech production to overcome potential ambiguities induced by T3 sandhi during an interactive task. Ten pairs of Chinese participants completed an interactive phrase matching task. Participants were shown displays with Chinese phrases (surname + title, e.g., 1a-b). One participant read an indicated phrase which the other selected from their display. There were two conditions. In the no sandhi condition (1a-b), the title did not induce sandhi, and should result in distinct F0 patterns for the surnames. In the sandhi condition (2a-b), the title induced T3 sandhi, and should result in homophonous surnames. 1a 卢侦探 (lu2 tʃ̺ən1tan4 “Detective Lu”) 2a 卢主任 (lu2 tʃ̺u3ɹ̺ən4 “Director Lu”) 1b 鲁侦探 (lu3 tʃ̺ən1tan4 “Detective Lu”) 2b 鲁主任 (lu3 tʃ̺u3ɹ̺ən4 “Director Lu”) Task performance showed clear evidence of sandhi-induced ambiguity. Examination of accuracy and tone acoustics suggested that pairs deployed different strategies to attempt to overcome sandhi-induced ambiguity, including exaggerating F0 rise or T3 duration, or adding a pause to avoid applying sandhi processes.

An Enhanced Phrase Matching Method Based on Cross‐Attention

Text matching is a core problem in the field of natural language understanding. It aims to analyze and judge the semantic relevance or similarity between two texts. In the past, much work on text matching focused on long text and little on phrase optimization. However, phrase matching also has essential application scenarios. Compared with long texts, phrases have difficulties covering less semantics and polysemy, and because of the weak expression ability of phrases, it is hard to match phrases accurately. On the Kaggle patent phrase matching dataset, due to the few words in the phrase and the repeated occurrences under different patent classification numbers, it is difficult to match each other accurately. In the data‐processing stage, this work proposes aggregating related targets for data fusion, expanding the background semantic information, and enhancing the expression ability of phrases. In the training stage, this work considers adding a cross‐attention network to the model to make the additional related targets better used and learned. The added cross‐attention network makes the model pay more attention to the most related information. The proposed method has experimented on pretrained language models such as BERT‐for‐patent, DeBERTa, and RoBERTa. The results show that the proposed method in this work improves by 2–4 points more than the general method without any information or additional network layers in its evaluation metric called the Pearson correlation, which enhances the performance in short‐text matching.

Research on Extraction Method of Financial Knowledge Based on How Net

Abstract In order to obtain the knowledge information of financial texts more efficiently and make the extracted information such as entity relation attribute more accurate, this paper studies the grammatical features of financial news texts and the semantic features of How Net, and puts forward the scheme of financial information extraction based on How Net. First, the phrase matching is carried out in the dictionary. Then the neural network is used for weighting, BiLSTM is used for character vector feature enhancement training, and then conditional random field (CRF) is used to complete named entity recognition, and then the relationship extraction of entity pairs from the dependency syntax is carried out to complete the research on the construction method of knowledge extraction of text in the financial field. The experimental results show that this model is superior to the other three models in entity recognition, and the overall performance is improved by about 1.2%. In relation extraction, the accuracy and recall rate of the model algorithm adopted in this paper are improved by 5% and 1.5% respectively, which shows that the improvement of the algorithm is effective.

An analysis of cognitive change in online mental health communities: A textual data analysis based on post replies of support seekers

The replies of people seeking support in online mental health communities can be analyzed to discover if they feel better after receiving support; feeling better indicates a cognitive change. Most research uses key phrase matching and word frequency statistics to identify psychological cognitive change, methods that result in omissions and inaccuracy. This study constructs an intelligent method for identifying psychological cognitive change based on natural language processing technology. It incorporates information related to emotions that appears in reply text to help identify whether psychological cognitive change has occurred. The model first encodes the emotion information based on rule matching and manual annotation, then adds the encoded emotion lexicon and a cognitive change lexicon to a word2vec high-dimensional semantic word vector training, converts the annotated cognitive change recognition text into a vector matrix using the trained model, and train in the annotated text using TextCNN. To compare the results with those of the traditional methods (key phrase matching and sentiment word frequency statistics), this study uses a semi-automated approach to construct a lexicon of psychological cognitive change, as well as a keyword lexicon without cognitive change, based on word vectors and similarity. We compare the performance of the classifier before and after the fusion of the graphical emotion information, compare the LSTM and Transformer as baselines, and compare traditional word frequency statistics methods. The experimental results show that our proposed classification model performs better than the others; it achieves 84.38% precision, an 84.09% recall rate, and an 84.17% F1 value. Our work bears methodological implications for online mental health platforms.

10,589 CT pulmonary angiograms: evaluating the yield of acute pulmonary embolism.

To evaluate the change in the number of CT pulmonary angiograms (CTPAs) performed and the change in the yield of acute pulmonary embolism (PE) on CTPA at a busy tertiary teaching hospital from 2016 to 2019. All CTPA examinations for both in-patients and emergency department patients performed at our busy tertiary teaching hospital between 1 January 2016 and 31 December 2019 were identified from the radiology information system. A natural language processing technique called phrase matching was employed to assign each of the examination reports a result of either positive, negative or equivocal for acute PE. This algorithm was validated on a sample of 200 reports. The number of CTPAs performed increased 59% from 2016 to 2019. The overall yield of acute PE has remained steady averaging 15.9%, ranging from 15.0% to 17.2%. Over 3 years, there has been a significant increase in the demand for CTPA examinations. The yield of acute PE has remained steady indicating a justified increase in demand. The yield of acute PE on CTPA within our centre is higher than the Royal College of Radiologists' suggested minimum of 15.4% which suggests the current guidelines used for the investigation of suspected acute PE within our centre are appropriate. The guidelines and subsequent yield of acute PE on CTPA at our tertiary teaching hospital can be used as a reference standard for other similar institutes.

Human-Computer Interaction Environment Monitoring and Collaborative Translation Mode Exploration Using Artificial Intelligence Technology.

Artificial intelligence now plays a significant role in both daily life and scientific research because of the rapid advancement of this technology in recent years. Making full use of the phrases in the translation phrase table for translation is challenging since the phrase matching is too accurate when the translation machine decodes. Fully automatic machine translation struggles to meet the expectations of its users since there are more or less translation faults brought on by data bottlenecks. Therefore, we require collaborative assisted translation technology for human-computer interaction. This work strengthens the research on collaborative translation techniques and ways for monitoring the human-computer interaction environment in order to further improve translation quality. This essay investigates and discusses human-computer translation techniques as well as related ideas in collaborative translation and human-computer interaction. The translation similarity model is incorporated into the translation system model together with an overall qualitative knowledge and logical reasoning capability of human-computer interaction to offer fresh strategies and methods for collaborative translation between humans and computers. According to the experimental findings, the accuracy rate of the collaborative translation system for human-computer interaction based on artificial intelligence technology can achieve 98.2% and 95.6%. The quality of the translation is enhanced after human-computer interaction, and the editing gap between the incorrect and auxiliary translations is narrowed, demonstrating the efficiency of the system and demonstrating its viability. In order to enhance the accuracy of system translation and the effectiveness of system operation, it is important to investigate the collaborative translation mode of human-computer interaction based on artificial intelligence technology.

Using Machine Learning to Detect Events on the Basis of Bengali and Banglish Facebook Posts

In modern times, ensuring social security has become the prime concern for security administrators. The widespread and recurrent use of social media sites is creating a huge risk for the lives of the general people, as these sites are frequently becoming potential sources of the organization of various types of immoral events. For protecting society from these dangers, a prior detection system which can effectively detect events by analyzing these social media data is essential. However, automating the process of event detection has been difficult, as existing processes must account for diverse writing styles, languages, dialects, post lengths, and et cetera. To overcome these difficulties, we developed an effective model for detecting events, which, for our purposes, were classified as either protesting, celebrating, religious, or neutral, using Bengali and Banglish Facebook posts. At first, the collected posts’ text were processed for language detection, and then, detected posts were pre-processed using stopwords removal and tokenization. Features were then extracted from these pre-processed texts using three sub-processes: filtering, phrase matching of specific events, and sentiment analysis. The collected features were ultimately used to train our Bernoulli Naive Bayes classification model, which was capable of detecting events with 90.41% accuracy (for Bengali-language posts) and 70% (for the Banglish-form posts). For evaluating the effectiveness of our proposed model more precisely, we compared it with two other classifiers: Support Vector Machine and Decision Tree.

The hybridised indexing method for research-based information retrieval

An information retrieval system (IRS) is used to retrieve documents based on an information need. The IRS makes relevance judgements by attempting to match a query to a document. As IRS capabilities are indexing design dependent, the hybrid indexing method (IRS-H) is introduced. The objectives of this article are to examine IRS-H (as an alternative indexing method that performs exact phrase matching) and IRS-I, regarding retrieval usefulness, identification of relevant documents, and the quality of rejecting irrelevant documents by conducting three experiments and by analysing the related data. Three experiments took place where a collection of 100 research documents and 75 queries were presented to: (1) five participants answering a questionnaire, (2) IRS-I to generate data and (3) IRS-H to generate data. The data generated during the experiments were statistically analysed using the performance measurements of Precision, Recall and Specificity, and one-tailed Student’s t-tests. The results reveal that IRS-H (1) increased the retrieval of relevant documents, (2) reduced incorrect identification of relevant documents and (3) increased the quality of rejecting irrelevant documents. The research found that the hybrid indexing method, using a small closed document collection of a hundred documents, produced the required outputs and that it may be used as an alternative IRS indexing method.

Multi stratagem analysis of sentiments on twitter data using partial phrase harmonizing

Sentiment analysis is constructive in the application environment for business intelligence and suggests systems because it is a very easy medium for the two ends of the availability to communicate. Numerous strategies and schemes have been worn inside the sentiment analysis, such as language processing, polarity lexicons, machine learning, and psychometric scales which establish diverse types of analyzing sentiments as assumptions ended, scheme reveals, and corroboration data set. Since the internet has to turn into a commanding resource of retrospect the sphere of sentiment is moreover referred to as Sentiment Analysis or Opinion Mining. It has seen an enormous boost in academia over the decades. Analyzing sentiment to extract sentiments in different levels like word, sentence, and document provides articles’ feeling polarities. While well identified consumers’ sentiments articulated in sentences by opinion. Customary machine learning schemes cannot virtuously mirror the views of writers. This paper proposes a scheme called multi-strategy sentiments with semantic resemblance to disentangle the topic with partial phrase matching. Additionally, the Naïve Bayes classification is applied to search for the probability of the distribution of knowledge in different categories of knowledge set.

Contextual advertising is an effective tool of digital marketing

The concept of contextual advertising, its strengths, main types are defined. Types of advertising campaigns are defined. Highlighted the basic settings in Google Ads, in particular the rules of keyword selection. The effectiveness of using contextual advertising as an Internet marketing tool has been established, given that it: acts selectively and is displayed to visitors of the website whose sphere of interests potentially coincides / intersects with the subject of the advertised product or service; meets the requirements of the target audience, which increases the likelihood of its response to advertising and makes it more attractive than other types of advertising; helps to improve the company's image, disseminate information about products and sell goods over the Internet; is an effective element of digital marketing, which with a high-quality setting can increase company revenues. Contextual advertising has prospects, given: - increasing the number of Internet users and the time of its use by each user; - its regional orientation, which allows you to weed out a lot of non-target customers; - focus on a specific target audience by interests; - ofline advertising is becoming more expensive; - contextual advertising tools are improving, more services are appearing for a deeper analysis of statistics. When setting up contextual advertising, the following are important: choosing the type of advertising campaign (image, product, trade), searching for keywords and choosing the type of correspondence (general matching, broad match modifier, phrase matching, exact matching). Effective contextual advertising is a chain of "request -> keyword -> advertisement -> landing page of the site-> purchase of goods". Keywords: digital marketing, search engine, contextual advertising, advertising campaign, demand, keywords.

Critical Examination of the Intertextual Phrase Matching Module in the Thesaurus Linguae Graecae and Its Relevance for Biblical and Patristic Studies

The rich availability of ancient Greek texts in the Thesaurus Linguae Graecae (TLG) has opened up new types of research for Biblical and Patristic scholars. A very helpful feature on the TLG’s website is the option to trace quotations with the help of an n-grams module (Intertextual Phrase Matching). However, it is virtually unknown how well this module performs and what scholars might expect from the results it produces. The core of this article, therefore, is devoted to a critical examination of the algorithm and of its results. The gospel according to John has been compared with the Paedagogus of Clement of Alexandria, with the gospel according to Matthew, and with the complete works of Plutarch. As it turns out, the algorithm performs well in cases of longer quotations with no or very few interpolations. Short quotations, however, are missed while interpolated or adapted quotations are poorly handled. It is suggested that the algorithm might perform better if the team of the TLG were to revisit its decision to ignore stopwords and if the algorithm were to allow for foreign words in its n-grams. Finally, it is advised that more transparency in the algorithm’s mechanisms and a possibility for manually tuning its parameters might improve its applicability.

A LONG DESIRED SELF-LEARNING SYSTEM TO AID BIG DATA BUSINESS SOCIAL MEDIA CONSUMPTION

ABSTRACT Organizations have developed Big Data capabilities to process enormous amounts of data – Petabytes and such. It is the lower end of the Volume-spectrum of Big Data that has not found enough easy to use, low infrastructure solutions that may require little learning yet may offer real time interaction possibilities and human friendly visualization. Typical media consumers such as airlines flyers have lagged in consumption of information, because Big Data business so far has failed to even acknowledge that the social media data average flyers can often have to deal with, has similar Big Data challenges. This experiment based research establishes and demonstrates a context-aware approach that can help flyers self-learn from blogs and their own social media data. It show-cases InContext, an in-browser Social Media data processor that has been tested with nearly 42,000 customer reviews expressed as nearly 0.5 million lines of text equivalent to nearly 17,500 A4 sized pages of text. InContext uses light-weight technologies for processing enormous amounts of Social Media data and carries out considerable Big Data tasks, real-time in an interactive manner without requiring the use of any (1) external dictionaries, (2) SQL or NoSQL like querying, and (3) libraries. InContext enables: (1) exploration of text data (2) keyword and phrase matching (3) data partitioning (4) data reduction (5) extraction of entities and dictionaries (6) extraction of relationships between entities (7) extraction of specific information in-context, say regarding food and drinks, in-flight entertainment or seating comfort and (8) comparison of flyer reviews with that specificity. This work has potential for significant impact on business. Especially the high usability and low resource requirements of InContext have implications for design of comparative apps by business. Also, by enabling flyers to consume more information – this work proposes to alleviate information load – thus increasing the availability of flyers to even more information from business. This work is novel, and we are not aware of any literature that addresses this issue. Keywords self-learning, context-aware, Big Data, ease of use, interactive, text visualization

Analysis of Short form Maintenance Records for NFF Using NLP, Phrase Matching, and Bayesian Learning

No Fault Found (NFF) is a well discussed phenomenon within the maintenance sector but which requires work to quantify how much of an issue it may be and provide metrics by which it may be tracked and various approaches to its reduction evaluated. Previous studies have relied on expert classification to identify NFF, however this approach is time consuming and costly. Maintainer classification (MC), expert classification (RC), phrase matching (PM), and Bayesian matching (NBPM) are all evaluated and contrasted as methods to identify NFF. The results demonstrate the utility of all 4 methods and discusses their place within a maintenance ecosystem.

EfficientPMM: Finite Automata Based Efficient Pattern Matching Machine

In text mining, vector space and bag of word models are poor candidates for topic discovery as they loose word order and co-occurrence, which are very crucial in understanding the meaning of a document. Phrase based models has proven to be promising in capturing the underlying document characteristics. This paper proposes a new document model and algorithm to perform efficient pattern matching for exact, prefix, postfix, and infix matching of phrases in near linear time. The pattern matching machine (PMM) uses the concepts of graph theory and the theory of automata to efficiently and intelligently match, index, track, and analyze phrase patterns. The scalability, space, and time performance are compared with the benchmark document models.

효율적인 한국어 파싱을 위한 최장일치 기반의 형태소 분석기 기능 확장

효율적인 한국어 파싱을 위한 최장일치 기반의 형태소 분석기 기능 확장

Utilising Wikipedia for Text Mining Applications

The process whereby inferences are made from textual data is broadly referred to as text mining. In order to ensure the quality and effectiveness of the derived inferences, several approaches have been proposed for different text mining applications. Among these applications, classifying a piece of text into pre-defined classes through the utilisation of training data falls into supervised approaches while arranging related documents or terms into clusters falls into unsupervised approaches. In both these approaches, processing is undertaken at the level of documents to make sense of text within those documents. Recent research efforts have begun exploring the role of knowledge bases in solving the various problems that arise in the domain of text mining. Of all the knowledge bases, Wikipedia on account of being one of the largest human-curated, online encyclopaedia has proven to be one of the most valuable resources in dealing with various problems in the domain of text mining. However, previous Wikipedia-based research efforts have not taken both Wikipedia categories and Wikipedia articles together as a source of information. This thesis serves as a first step in eliminating this gap and throughout the contributions made in this thesis, we have shown the effectiveness of Wikipedia category-article structure for various text mining tasks. Wikipedia categories are organized in a taxonomical manner serving as semantic tags for Wikipedia articles and this provides a strong abstraction and expressive mode of knowledge representation. In this thesis, we explore the effectiveness of this mode of Wikipedia's expression (i.e., the category-article structure) via its application in the domains of text classification, subjectivity analysis (via a notion of "perspective" in news search), and keyword extraction. First, we show the effectiveness of exploiting Wikipedia for two classification tasks i.e., 1- classifying the tweets1 being relevant/irrelevant to an entity or brand, 2- classifying the tweets into different topical dimensions such as tweets related with workplace, innovation, etc. To do so, we define the notion of relatedness between the text in tweet and the information embedded within the Wikipedia category-article structure. Then, we present an application in the area of news search by using the same notion of relatedness to show more information related to each search result highlighting the amount perspective or subjective bias in each returned result towards a certain opinion, topical drift, etc. Finally, we present a keyword extraction strategy using community detection over the Wikipedia categories to discover related keywords arranged in different communities. The relationship between Wikipedia categories and articles is explored via a textual phrase matching framework whereby the starting point is textual phrases that match Wikipedia articles' titles/redirects. The Wikipedia articles for which a match occurs are then utilised by extraction of their associated categories, and these Wikipedia categories are used to derive various structural measures such as those relating to taxonomical depth and Wikipedia articles they contain. These measures are utilised in our proposed text classification, subjectivity analysis, and keyword extraction framework and the performance is analysed via extensive experimental evaluations. These experimental evaluations undertake comparisons with standard text mining approaches in the literature and our Wikipedia framework based on its category-article structure outperforms the standard text mining techniques.

Bayes pooling of visual phrases for object retrieval

Object retrieval is still an open question. A promising approach is based on the matching of visual phrases. However, this routine is often corrupted by visual phrase burstiness, i.e., the repetitive occurrence of some certain visual phrases. Burstiness leads to over-counting the co-occurring visual patterns between two images, thus would deteriorate the accuracy of image similarity measurement. On the other hand, existing methods are incapable of capturing the complete geometric variation between images. In this paper, we propose a novel strategy to address the two problems. Firstly, we propose a unified framework for matching geometry-constrained visual phrases. This framework provides a possibility of combing the optimal geometry constraints to improve the validity of matched visual phrases. Secondly, we propose to address the problem of visual phrase burstiness from a probabilistic view. This approach effectively filters out the bursty visual phrases through explicitly modelling their distribution. Experiments on five benchmark datasets demonstrate that our method outperforms other approaches consistently and significantly.

Adapted Web Crawler for Mining Offline Web Data using AFHC

focused hyperlink crawler (AFHC) aim to search the entire inner level sub link of the web pages related to a specific topic and to download unique web pages to local disk. The download web page information searched in offline browsing and avoids the repeated searches in the web server to give a solution to problem. The major problem is to retrieve the maximal set of relevant and quality pages. Crawler software can be retrieve web pages by hyperlinks through internet. The number of web sites using web crawlers cannot retrieve the relevant pages. The system have browser and search engine. Browser using AFHC reduce time to finding accurate content in web pages in the hyperlinks and also restrict to download web pages to local disk. Search engine using extended cocitaion algorithm to retrieve accurate content in the local disk and search based on any word, all word and phrase matching in the local disk. It is useful to the student and organization. The research is easily found to personalize the crawl history and search history for knowing the user transactions.

Assessing the use of multiple sources in student essays

The present study explored different approaches for automatically scoring student essays that were written on the basis of multiple texts. Specifically, these approaches were developed to classify whether or not important elements of the texts were present in the essays. The first was a simple pattern-matching approach called "multi-word" that allowed for flexible matching of words and phrases in the sentences. The second technique was latent semantic analysis (LSA), which was used to compare student sentences to original source sentences using its high-dimensional vector-based representation. Finally, the third was a machine-learning technique, support vector machines, which learned a classification scheme from the corpus. The results of the study suggested that the LSA-based system was superior for detecting the presence of explicit content from the texts, but the multi-word pattern-matching approach was better for detecting inferences outside or across texts. These results suggest that the best approach for analyzing essays of this nature should draw upon multiple natural language processing approaches.