Search

In this study, based on a morpheme segmentation framework, we researched a text keyword extraction method for Uyghur, Kazakh and Kirghiz languages, which have similar grammatical and lexical structures. In these languages, affixes and a stem are joined together to form a word. A stem is a word particle with a notional meaning, while the affixes perform grammatical functions. Because of these derivative properties, the vocabularies used for these languages are huge. Therefore, pre-processing is a necessary step in NLP tasks for Uyghur, Kazakh and Kirghiz. Morpheme segmentation enabled us to remove the suffixes as the auxiliary unit while retaining the meaningful stem and it reduced the dimension of the feature space present in the keyword extraction task for Uyghur, Kazakh and Kirghiz texts. We transformed the morpheme segmentation task into the problem of labeling the morpheme sequences, and we used the Bi-LSTM network to bidirectionally obtain the position feature information of character sequences. We applied CRF to effectively learn the information of the preceding and following label sequences to build a highly accurate Bi-LSTM_CRF morpheme segmentation model, and we prepared morpheme-based experimental text sets by using this model. Subsequently, we used the stem vectors’ similarity to modify the TextRank algorithm, subsequent to the training of the stem embedding vector using the Doc2vec algorithm, and then we performed a text keyword extraction experiment. In this experiment, the highest F1 scores of 43.8%, 44% and 43.9% were obtained for three datasets. The experimental results show that the morpheme-based approach provides much better results than the word-based approach, which shows the stem vector similarity weighting is an efficient method for the text keyword extraction task, thus proving the efficiency of morpheme sequence for morphologically derivative languages.

Aiming at the shortcomings of the TextRank method (TM) which only considers the co-occurrence between words and the incipient word importance when extracting keywords, this paper proposes a tolerance rough set (TRS)-based unsupervised keyword extraction method. Generally, how to score the words in a document has a significant influence on the word graph modeling. In this paper, we improve TM in two aspects with TRS theory that is used to mine vocabulary, semantics, grammar and other information in the corpus. First, the degree of words belonging to each document is calculated to form a fuzzy membership matrix, which helps to characterize the incipient word importance. Second, the fuzzy membership of words to each word tolerance class is calculated to form a semantic correlation matrix, which contributes to optimize the transition probability of all graph edges. We apply the proposed methods to the clustering tasks of two datasets, outperforming the strong baselines. [ABSTRACT FROM AUTHOR]

Increasingly, the amount of data in the form of text documents scattered on the internet is getting bigger. It took a very long time to get the information from each of these documents. For this reason, several researchers developed the Automatic Text Summarizer to summarize text automatically, so that the time needed to get important information from the entire document can be faster. Research that focuses on automatic summarization of Arabic texts is very rare. In fact, there are more than 300 million Arabic speakers in the world and Arabic is the official language at the United Nations. Therefore, this study develops a model that can perform text summarization automatically using the TextRank algorithm. The test results using Q&A Evaluation show very good results with details of the suitability of the summary results with the original text by 90%, the suitability of the summary results with Arabic grammar is 91.43%, the suitability of the summary results is 90%, the ease of understanding the summary results is 90%. and the useful aspects of the model developed were 91.43%.

In the era of rapid development of social networks, user reviews of restaurant review websites have grown rapidly. In order to allow users to quickly grasp the key points of review information on review sites, this paper provides an abstractive multi-text summary method that can automatically generate template-based review summaries based on predefined topics and sentiments. In particular, for each predefined topic and each type of sentiment (positive or negative), this study uses the TextRank algorithm to find the most representative sentences to form a summary. This method allows users to quickly grasp the positive and negative opinions of each important aspect of the restaurant. The previous research on generating abstracts from reviews either did not generate abstracts based on topics, or they were based on topics generated by random models. However, the latter method cannot guarantee that the topics generated by the random model are really the topics that the user needs. For a restaurant review, some topics are indispensable. In order to ensure that abstracts can be generated for these essential topics, our method predefines the topics that must be generated, and then generates abstracts for these topics. In the evaluation, this study compared the template method with the Refresh and Gensim systems based on criteria such as informativeness, clarity, usefulness and likes. The results show that the method proposed in this paper is superior to the other two summary methods.

Release notes are admitted as an essential technical document in software maintenance. They summarize the main changes, e.g. bug fixes and new features, that have happened in the software since the previous release. Manually producing release notes is a time-consuming and challenging task. For that reason, sometimes developers neglect to write release notes. For example, we collect data from GitHub with over 1900 releases, and among them, 37% of the release notes are empty. To mitigate this problem, we propose an automatic release notes generation approach by applying the text summarization techniques, i.e. TextRank. To improve the keyword extraction method of traditional TextRank, we integrate the GloVe word embedding technique with TextRank. After generating release notes automatically, we apply machine learning algorithms to classify the release note contents (or sentences). We classify the contents into six categories, e.g. bug fixes and performance improvements, to represent the release notes better for users. We use the evaluation metric, e.g. ROUGE, to evaluate the automatically generated release notes. We also compare the performance of our technique with two popular extractive algorithms, e.g. Luhn's and latent semantic analysis (LSA). Our evaluation results show that the improved TextRank method outperforms the two algorithms. [ABSTRACT FROM AUTHOR]

Automatic keywords extraction is a method that extracts words or phrases from a document which can express the main idea of the document. In this paper, we propose an unsupervised keywords extraction framework for individual documents, which improves the keywords extraction from two aspects. In the step of candidate keywords selection, we use the methods of removing the stopwords, regular matching, and length filtering to reduce the number of candidate keywords, but improve the quality. In the step of scoring words, we use word co-occurrence, semantic relationships (WordNet, Word Embedding, Normalized Google Distance), and three ways to combine word co-occurrence and semantic relationships to measure the weight of edges in the graph model. In experiments, we use Precision, Recall, and F1-measure values as evaluation criteria to compare all keywords extraction methods we proposed with other strong baseline methods in two datasets. According to the results of experiments, methods under our proposed framework achieve good results. We verify that the methods of using both word co-occurrence and semantic relationships have a better effect on keywords extraction than using co-occurrence or semantic relationships only. At the same time, we also find that for the keywords extraction of individual documents, the method of using co-occurrence between words has a better effect than semantic relationships.

As a typical keyword extraction technology, TextRank has been used in a wide variety of commercial applications, including text classification, information retrieval and clustering. In these applications, the parameters of TextRank, including the co-occurrence window size, iteration number and decay factor, are set roughly, which might affect the effectiveness of returned results. In this work, we conduct an empirical study on TextRank, towards finding optimal parameter settings for keyword extraction. The experiments are done in Hulth2003 and Krapivin2009 datasets, which are two real datasets. We first remove the stop word by an open published English stop word list XPO6. And then, we extract the word stems by Porter Stemmer. Porter Stemmer is a tool which can find the stems of words with multiple variants, discard redundant information, strengthen the filtering effect, and extract the effective features of the text fully. We carry out extensive experiments to evaluate the effects of the parameters to keywords extraction, and evaluate the effectiveness of corresponding results by Precision, Recall and Accuracy. Experimental results show that TextRank shows the best performance when setting co-occurrence window size w = 3, iteration number t = 20, decay factor c = 0.9 and rank k = 10 respectively, and the results are independent of the text length.

Abstract As the amount of published geoscience literature grows, reading and summarizing texts of large collections has become a challenging task. Publication keywords can be considered basic components of knowledge structure representations and have been used to reveal knowledge concerning research domains. In contrast to data used in other research domains, the works on textual geoscience data that entail keyword extraction are limited. In this paper, we propose an unsupervised algorithm, the graph‐based keyword extractor with error‐feedback propagation (GKEEP), that enhances graph‐based keyword extraction approaches by using an error‐feedback mechanism similar to the concept of backpropagation. The proposed approach comprises the following steps. A preprocessed document is used as the input of the proposed model and is represented as a weighted undirected graph, where the vertices represent words and the edges represent the cooccurrence relationship between the words constrained by a window size. Subsequently, its nodes are ranked by their importance scores calculated by a graph‐based ranking algorithm. Consequently, all the words have their own scores, and they are used to compute the scores of keyword candidates. Subsequently, the Word2Vec method is applied to recalculate the scores of keyword candidates and rank the keyword candidates to select the final keyword. It also utilizes error feedback to boost the rankings of the most salient terms that would otherwise be deemed less important. With empirical experiments on two real data sets (including our newly built data set), the proposed GKEEP model outperforms state‐of‐the‐art unsupervised models and the existing graph‐based ranking models. The proposed method can effectively reflect intrinsic keyword semantics and interrelationships.

With the development of information technology such as mobile Internet and social media applications, network information is growing rapidly and leads to the problem of information overload. Keywords help to filter and find interesting information for users from massive text. Automatic extraction of keywords from text as tags of text help to improve recommendation and keyword-based information retrieval. This paper proposes a novel keyword extraction approach from text that combines features such as word frequency and association. Experiment results show that the precision rate, recall rate and F-measure are all better than those of TextRank and TF-IDF. [ABSTRACT FROM AUTHOR]

As the amount of published geoscience literature grows, reading and summarizing texts of large collections has become a challenging task. Publication keywords can be considered basic components of knowledge structure representations and have been used to reveal knowledge concerning research domains. In contrast to data used in other research domains, the works on textual geoscience data that entail keyword extraction are limited. In this paper, we propose an unsupervised algorithm, the graph‐based keyword extractor with error‐feedback propagation (GKEEP), that enhances graph‐based keyword extraction approaches by using an error‐feedback mechanism similar to the concept of backpropagation. The proposed approach comprises the following steps. A preprocessed document is used as the input of the proposed model and is represented as a weighted undirected graph, where the vertices represent words and the edges represent the cooccurrence relationship between the words constrained by a window size. Subsequently, its nodes are ranked by their importance scores calculated by a graph‐based ranking algorithm. Consequently, all the words have their own scores, and they are used to compute the scores of keyword candidates. Subsequently, the Word2Vec method is applied to recalculate the scores of keyword candidates and rank the keyword candidates to select the final keyword. It also utilizes error feedback to boost the rankings of the most salient terms that would otherwise be deemed less important. With empirical experiments on two real data sets (including our newly built data set), the proposed GKEEP model outperforms state‐of‐the‐art unsupervised models and the existing graph‐based ranking models. The proposed method can effectively reflect intrinsic keyword semantics and interrelationships. Plain Language Summary: The common or frequently used terms receive higher scores in traditional graph‐based extraction owing to there are more edges connected to them. This paper proposes a graph‐based KE algorithm called KE using error‐feedback propagation, which utilizes the semantics of word embedding to assist in extracting keywords from geoscience reports. We hope that our approach will serve as an alternative method that deserves further study. Key Points: Word embedding is incorporated to capture the dependency structure as well as the data distribution, and it computes semantic relations to solve the content sparsity problemError feedback is utilized to boost the most salient terms that graph‐based approaches deem less importantA set of experiments to verify the effectiveness of the proposed method on two available manually constructed data sets [ABSTRACT FROM AUTHOR]

In the fast growing of digital technologies, crawlers and search engines face unpredictable challenges. Focused web-crawlers are essential for mining the boundless data available on the internet. Web-Crawlers face indeterminate latency problem due to differences in their response time. The proposed work attempts to optimize the designing and implementation of Focused Web-Crawlers using Master-Slave architecture for Bioinformatics web sources. Focused Crawlers ideally should crawl only relevant pages, but the relevance of the page can only be estimated after crawling the genomics pages. A solution for predicting the page relevance, which is based on Natural Language Processing, is proposed in the paper. The frequency of the keywords on the top ranked sentences of the page determines the relevance of the pages within genomics sources. The proposed solution uses a TextRank algorithm to rank the sentences, as well as ensuring the correct classification of Bioinformatics web page. Finally, the model is validated by being compared with a breadth first search web-crawler. The comparison shows significant reduction in run time for the same harvest rate.

Railways have developed rapidly in China for several decades. The hardware of railways has already reached the world's leading level, but the level of service of these railways still has room for improvement. The railway management department receives a large number of passenger complaints every year and records them in text, which needs to be classified and analyzed. The text of railway complaints includes characteristics spanning wide business coverage, various events, serious colloquialisms, interference and useless information. When using the direct classification via traditional text categorization, the classification accuracy is low. The key to the automatic classification of such text lies in an eigenvalue extraction. The more accurate the eigenvalue extraction, the higher the accuracy of text classification. In this paper, the TF-IDF algorithm, TextRank algorithm and Word2vec algorithm are selected to extract text eigenvalues, and a railway complaint text classification method is constructed with a naive Bayesian classifier. The three types of eigenvalue extraction algorithms are compared. The TF-IDF algorithm, based on eigenvalue extraction, achieves the highest automatic text classification accuracy.

Aiming at the traditional graph models for text summarization only focus on statistical features or shallow semantic features, and lack mining and utilization of deep topic semantic features, this paper proposed MDSR( multi-dimension summarization rank), an automatic text summarization method that combined topic feature. Specifically, this method adopted the LDA model to mine the semantic information of text topics and measured the impact of topic feature on a sentence by defining the importance of the topic. And it improved the construction mode of the probability transition matrix of graph model nodes by combining the topic feature with statistic features and inter-sentence similarity. Finally, it extracted and measured summarization according to the weight of sentence nodes. The results show that the ROUGE value evaluates by MDSR reaches the best when the weight ratio of topic feature, statistic feature and inter-sentence similarity is 3 : 4 : 3. The ROUGE-1, ROUGE-2, ROUGE-SU4 are 53. 35%,35. 18% and 33. 86%, which perform better than other comparisons. It shows that the text summarization method combining topic feature can effectively improve the accuracy of the summarization extraction. [ABSTRACT FROM AUTHOR]

This work presents FuzzyFeatureRank, a new feature reduction method inspired on PageRank to reduce the dimensionality of the feature space in supervised classification problems. More precisely, as it relies on a weighted directed graph, it is ultimately inspired on TextRank, a PageRank based method that adds weights to the edges to express the strength of the connections between nodes. The method is based on dividing each original feature used to describe the data into a set of fuzzy predicates and then ranking all of them by their ability to differentiate among classes in the light of the training set. In order to do that, both the information gained by each predicate and their redundancy with other already selected predicates are taken into account. The fuzzy predicates with the best scores can then be used as a reduced input to construct fuzzy classifiers that consider only the preselected predicates to build the antecedents of the fuzzy rules. The novelty of the proposal relies on being an approach halfway between feature selection and feature extraction approaches, being able to improve the discrimination ability of the original features but preserving the interpretability of the new features in the sense that they are fuzzy expressions. The experimental results support the suitability of the proposal. [ABSTRACT FROM AUTHOR]

Wikipedia is one of the main sources of information on the Web. But the access to this content may be difficult especiallywhen using a basic telephone without browsing capability and only a GSM network. The only means of text-basedcommunication remains through SMS. Due to the limitation of the number of characters, a Wikipedia page cannot alwaysbe sent through SMS. This work raises the issue of text summarization with character limitation. To solve this issue, twoextractive approaches have been combined: LSA and TextRank algorithms. Generated summaries have been evaluated usingROUGE metrics. Since ROUGE metrics do not consider character limitation, a new threshold named Threshold ofAcceptability for Character-Oriented Summaries (TACOS) has been proposed to appreciate ROUGE metrics. The evaluationshowed the relevance of the approach for pages of at most 2000 characters. The system has been tested using the SMSsimulator of RapidSMS without a GSM gateway to simulate the deployment in a real environment. To the best of ourknowledge, this is the first work tackling text summarization issue with character limitation.

Wikipedia is one of the main sources of information on the Web. But the access to this content may be difficult especially when using a basic telephone without browsing capability and only a GSM network. The only means of text-based communication remains through SMS. Due to the limitation of the number of characters, a Wikipedia page cannot always be sent through SMS. This work raises the issue of text summarization with character limitation. To solve this issue, two extractive approaches have been combined: LSA and TextRank algorithms. Generated summaries have been evaluated using ROUGE metrics. Since ROUGE metrics do not consider character limitation, a new threshold named Threshold of Acceptability for Character-Oriented Summaries (TACOS) has been proposed to appreciate ROUGE metrics. The evaluation showed the relevance of the approach for pages of at most 2000 characters. The system has been tested using the SMS simulator of RapidSMS without a GSM gateway to simulate the deployment in a real environment. To the best of our knowledge, this is the first work tackling text summarization issue with character limitation. [ABSTRACT FROM AUTHOR]

Topic word extraction is the task of identifying single or multi-word expressions that represent the main topics of a document. In this paper, two improved algorithms for extracting and discovering topic words are proposed in the Rapid Topic word Detection (RTD) Algorithm and CategoryTextRank (CTextRank) Algorithm, which can effectively obtain information by extracting and filtering the topic words in the text. The algorithms overcome the shortcomings of traditional topic words discovering algorithms that require deep linguistic knowledge, domain or language specific annotated corpora. The two algorithms we proposed can process both short and long text. The biggest advantage of the algorithms is that they are unsupervised machine learning algorithms. They need not be trained to process text directly to get topic words. The Accuracy rate, recall rate and F-measure index have been greatly improved when using the two algorithms which show that the results obtained compare favorably with previously published results on datasets Inspec and SemEval. The first algorithm Rapid Topicword Detection improves the metrics compared to PositionRank and TextRank, the second algorithm CategoryTextRank improves the metrics compared to TextRank, SingleRank and TF-IDF. [ABSTRACT FROM AUTHOR]

Studying the Qur'an by understanding the Qur'an's vocabulary so that understanding its meaning is not easy. Then we need a Qur'anic vocabulary encyclopedia that focuses on explaining the meaning of the words in it. The development of encyclopedias used automatic text summarization with the TextRank method because from one query that is searched there are many meanings of words that must be summarized. This method starts by selecting documents that are relevant to the query, then summarizes the selected documents using the TextRank method so that they get a summary based on all the word meanings, finally testing the summary results by the system by comparing them with summary targets that are constructed manually by humans. The application of the TextRank method for automatic text summarization has an average value of F-Score 0.6173. The results of automatic text summarization using the TextRank method are no duplicates, and for some queries, it is almost the same as the summary results created manually by humans. [ABSTRACT FROM AUTHOR]

Jacob is a voice chatbot that use Wit.ai to get the context of the question and give an answer based on that context. However, Jacob has no variation in answer and could not recognize the context well if it has not been learned previously by the Wit.ai. Thus, this paper proposes two features of artificial intelligence (AI) built as a web service: the paraphrase of answers using the Stacked Residual LSTM model and the question summarization using Cosine Similarity with pre-trained Word2Vec and TextRank algorithm. These two features are novel designs that are tailored to Jacob, this AI module is called Cleveree. The evaluation of Cleveree is carried out using the technology acceptance model (TAM) method and interview with Jacob admins. The results show that 79.17% of respondents strongly agree that both features are useful and 72.57% of respondents strongly agree that both features are easy to use. [ABSTRACT FROM AUTHOR]

The method of Chinese existing automatic text summarization mainly utilizes the text's own information, and its defect is that it cannot make full use of the related semantic information between the words. Therefore, this paper proposed an improved Chinese text summarization method. This method integrated the information of the external corpora into the TextRank algorithm in the form of a word vector. Combined TextRank with word2vec, it mapped each word in the sentence to the high dimensional lexicon to form a sentence vector. This method fully considered the similarity between sentences, the coverage of keywords and the similarity between sentence and title to calculate the influence weights among sentences, and choose the top-ranked sentences used as the summarization of the text. The results of experiment show that this method achieves good results in the data set of this paper, and is more effective than the original method in extracting Chinese summarization automatically. [ABSTRACT FROM AUTHOR]

Actual problem in nowadays is to efficiently process the large amount of data that pass through our mind everyday. The object of study of this paper is automatic summarization algorithms. The main goal is to implement and make comparison of different summarization techniques on corpora of news articles parsed from the web. This research work contains the description of three summarization techniques based on TextRank algorithm: General TextRank, BM25, LongestCommonSubstring. It is specially noted the languages of used corpora: Russian and Kazakh languages. The results of summarization processes and their comparison are provided. It should be emphasized that used algorithms are well-known, but the way of their evaluation on defined corpora is different from those which usually used in summary evaluation. The method of summary evaluation proposed use the special dictionary of extracted key-words on the topic of corpora. As the title implies the article describes applying statistical information. The semantic and syntactic features of text are not examined.

Automatic extractive text summarization retrieves a subset of data that represents most notable sentences in the entire document. In the era of digital explosion, which is mostly unstructured textual data, there is a demand for users to understand the huge amount of text in a short time; this demands the need for an automatic text summarizer. From summaries, the users get the idea of the entire content of the document and can decide whether to read the entire document or not. This work mainly focuses on generating a summary from multiple news documents. In this case, the summary helps to reduce the redundant news from the different newspapers. A multi-document summary is more challenging than a single-document summary since it has to solve the problem of overlapping information among sentences from different documents. Extractive text summarization yields the sensitive part of the document by neglecting the irrelevant and redundant sentences. In this paper, we propose a framework for extracting a summary from multiple documents in the Malayalam Language. Also, since the multi-document summarization data set is sparse, methods based on deep learning are difficult to apply. The proposed work discusses the performance of existing standard algorithms in multi-document summarization of the Malayalam Language. We propose a sentence extraction algorithm that selects the top ranked sentences with maximum diversity. The system is found to perform well in terms of precision, recall, and F-measure on multiple input documents.

The rapid development of social media encourages people to share their opinions and feelings on the Internet. Every day, a large number of short text comments are generated through Twitter, microblogging, WeChat, etc., and there is high commercial and social value in extracting useful information from these short texts. At present, most studies have focused on extracting text key words. For example, the LDA topic model has good performance with long texts, but it loses effectiveness with short texts because of the noise and sparsity problems. In this paper, we attempt to use Word2Vec and Doc2Vec to improve short-text key word extraction. We first added the method of the collaborative training of word vectors and paragraph vectors and then used the TextRank model's clustering nodes. We adjusted the weights of the key words that were generated by computing the jump probability between nodes and then obtained the node-weighted score, and eventually sorted the generated key words. The experimental results show that the improved method has good performance on the datasets. [ABSTRACT FROM AUTHOR]

This paper studied the influence of lexical semantic difference on TextRank algorithm. and presented a keyword extraction method based on word vector and TextRank. Firstly.it used FastText to represent word vector from the document corpus. Then,based on the idea of implicit subject distribution and used the differences in lexical semantics to build a probability transfer matrix for TextRank. Finally.it iteratively calculated the lexical graph model and extracted keywords. Experimental results show that the extraction performance of this method is significantly improved compared with the traditional method. In addition, it is proved that the use of word vectors can improve the performance of TextRank algorithm simply and effectively. [ABSTRACT FROM AUTHOR]

When abstracting Chinese texts, the traditional TextRank algorithm only considers the similarity between nodes and neglects other important information of the text. Firstly, aiming at Chinese single document, on the basis of existing research, this paper used TextRank algorithm,on the one hand, it considered the similarities between sentences, on the other hand, Text-Rank was combined with the overall structural information of texts and the contextual information of sentences, such as the physical position of the document sentences or paragraph, feature sentences, core sentences and other sentences that might increase the weight of the sentence, all were used to generate the digest candidate sentence group of the text. And then, removing high-similarity sentences by redundancy processing technology on the digest candidate sentence group. Finally, the experimental verification shows that the algorithm can improve the accuracy of the generated digest, indicating the effectiveness of the algorithm. [ABSTRACT FROM AUTHOR]

Automatic question generation from text or paragraph is a great challenging task which attracts broad attention in natural language processing. Because of the verbose texts and fragile ranking methods, the quality of top generated questions is poor. In this paper, we present a novel framework Automatic Chinese Question Generation (ACQG) to generate questions from text or paragraph. In ACQG, we use an adopted TextRank to extract key sentences and a template-based method to construct questions from key sentences. Then a multi-feature neural network model is built for ranking to obtain the top questions. The automatic evaluation result reveals that the proposed framework outperforms the state-of-the-art systems in terms of perplexity. In human evaluation, questions generated by ACQG rate a higher score. [ABSTRACT FROM AUTHOR]

As a classical key-word extracting and abstraction auto-generating algorithm, Text Rank considered the text as a group of terms, and sought a latent semantic relationship between terms according to iteratively calculating the weights of the terms in the nodes graph. Based on the nodes graph model of TextRank, combined node graph and Markov state transform model, weighted the edge between nodes with conditional probability, proposed a new nodes graph model and corresponding algorithm TextRank_Revised(TR-R). According to the verification on labeled and unlabeled samples, it shows that without promotion of time complexity, the new algorithm can get a key-word sorting consequence which is closer to the manual than the original algorithm from the single text. [ABSTRACT FROM AUTHOR]