Search

Feedback is important to improve writing quality; however, to provide timely and personalized feedback is a time-intensive task. Currently, most literature focuses on providing (human or machine) support on product characteristics, especially after a draft is submitted. However, this does not assist students who struggle during the writing process. Therefore, in this study, we investigate the use of keystroke analysis to predict writing quality throughout the writing process. Keystroke data were analyzed from 126 English as a second language learners performing a timed academic summarization task. Writing quality was measured using participants’ final grade. Based on previous literature, 54 keystroke features were extracted. Correlational analyses were conducted to identify the relationship between keystroke features and writing quality. Next, machine learning models (regression and classification) were used to predict final grade and classify students who might need support at several points during the writing process. The results show that, in contrast to previous work, the relationship between writing quality and keystroke data was rather limited. None of the regression models outperformed the baseline, and the classification models were only slightly better than the majority class baseline (highest AUC = 0.57). In addition, the relationship between keystroke features and writing quality changed throughout the course of the writing process. To conclude, the relationship between keystroke data and writing quality might be less clear than previously posited.

Short text clustering has far-reaching effects on semantic analysis, showing its importance for multiple applications such as corpus summarization and information retrieval. However, it inevitably encounters the severe sparsity of short text representations, making the previous clustering approaches still far from satisfactory. In this paper, we present a novel attentive representation learning model for shot text clustering, wherein cluster-level attention is proposed to capture the correlations between text representations and cluster representations. Relying on this, the representation learning and clustering for short texts are seamlessly integrated into a unified model. To further ensure robust model training for short texts, we apply adversarial training to the unsupervised clustering setting, by injecting perturbations into the cluster representations. The model parameters and perturbations are optimized alternately through a minimax game. Extensive experiments on four real-world short text datasets demonstrate the superiority of the proposed model over several strong competitors, verifying that robust adversarial training yields substantial performance gains., Comment: 14pages, to appear in IEEE TKDE

In this paper, we propose a novel deep Efficient Relational Sentence Ordering Network (referred to as ERSON) by leveraging pre-trained language model in both encoder and decoder architectures to strengthen the coherence modeling of the entire model. Specifically, we first introduce a divide-and-fuse BERT (referred to as DF-BERT), a new refactor of BERT network, where lower layers in the improved model encode each sentence in the paragraph independently, which are shared by different sentence pairs, and the higher layers learn the cross-attention between sentence pairs jointly. It enables us to capture the semantic concepts and contextual information between the sentences of the paragraph, while significantly reducing the runtime and memory consumption without sacrificing the model performance. Besides, a Relational Pointer Decoder (referred to as RPD) is developed, which utilizes the pre-trained Next Sentence Prediction (NSP) task of BERT to capture the useful relative ordering information between sentences to enhance the order predictions. In addition, a variety of knowledge distillation based losses are added as auxiliary supervision to further improve the ordering performance. The extensive evaluations on Sentence Ordering, Order Discrimination, and Multi-Document Summarization tasks show the superiority of ERSON to the state-of-the-art ordering methods.

Extracting relevant information from a large amount of data is a challenging task. Automatic text summarization is a potential solution for obtaining this information. In this paper, a nature inspired swarm intelligence-based algorithm viz. firefly algorithm for multi-document text summarization is proposed. A new fitness function consisting of three features viz. topic relation factor, cohesion factor and readability factor is utilized. The experiments are performed on datasets from Document Understanding Conference i.e. DUC-2002, DUC-2003 and DUC-2004. The performance of the algorithm has been evaluated using ROUGE score. The performance of the proposed algorithm is compared with some other nature inspired ones such as particle swarm optimization (PSO) and genetic algorithm (GA). The performance of the proposed algorithm outperforms the other adopted ones.

In the natural language processing family, learning representations is a pioneering study, especially in sequence-to-sequence tasks where outputs are generated, totally relying on the learning representations of source sequence. Generally, classic methods infer that each word occurring in the source sequence, having more or less influence on the target sequence, should all be considered when generating outputs. As the summarization task requires the output sequence to only retain the essence, classic full consideration of the source sequence may not work well on it, which calls for more suitable methods with the ability to discard the misleading noise words. Motivated by this, with both relevance retaining and redundancy removal in mind, we propose a summarization learning model by implementing an encoder with copious contextual information represented and a decoder with a selecting mechanism integrated. Specifically, we equip the encoder with an asynchronous bi directional parallel structure, in order to obtain abundant semantic representation. The decoder, different from the classic attention-based works, employs a self-aware context selecting mechanism to generate summary in a more productive way. We evaluate the proposed methods on three benchmark summarization corpora. The experimental results demonstrate the effectiveness and applicability of the proposed framework in relation to several well-practiced and state-of-the-art summarization methods.

The day-to-day growth of online information necessitates intensive research in automatic text summarization (ATS). The ATS software produces summary text by extracting important information from the original text. With the help of summaries, users can easily read and understand the documents of interest. Most of the approaches for ATS used only local properties of text. Moreover, the numerous properties make the sentence selection difficult and complicated. So this article uses a graph based summarization to utilize structural and global properties of text. It introduces maximal clique based sentence selection (MCBSS) algorithm to select important and non-redundant sentences that cover all concepts of the input text for summary. The MCBSS algorithm finds novel information using maximal cliques (MCs). The experimental results of recall oriented understudy for gisting evaluation (ROUGE) on Timeline dataset show that the proposed work outperforms the existing graph algorithms Bushy Path (BP), Aggregate Similarity (AS), and TextRank (TR).

In the era of social networks, the rapid growth of data mining in information retrieval and natural language processing makes automatic text summarization necessary. Currently, pretrained word embedding and sequence to sequence models can be effectively adapted in social network summarization to extract significant information with strong encoding capability. However, how to tackle the long text dependence and utilize the latent topic mapping has become an increasingly crucial challenge for these models. In this article, we propose a topic-aware extractive and abstractive summarization model named T-BERTSum, based on Bidirectional Encoder Representations from Transformers (BERTs). This is an improvement over previous models, in which the proposed approach can simultaneously infer topics and generate summarization from social texts. First, the encoded latent topic representation, through the neural topic model (NTM), is matched with the embedded representation of BERT, to guide the generation with the topic. Second, the long-term dependencies are learned through the transformer network to jointly explore topic inference and text summarization in an end-to-end manner. Third, the long short-term memory (LSTM) network layers are stacked on the extractive model to capture sequence timing information, and the effective information is further filtered on the abstractive model through a gated network. In addition, a two-stage extractive-abstractive model is constructed to share the information. Compared with the previous work, the proposed model T-BERTSum focuses on pretrained external knowledge and topic mining to capture more accurate contextual representations. Experimental results on the CNN/Daily mail and XSum datasets demonstrate that our proposed model achieves new state-of-the-art results while generating consistent topics compared with the most advanced method.

The availability of legal judgment documents in digital form offers numerous opportunities for information extraction and application. Automatic summarization of these legal texts is a crucial and a challenging task due to the unusual structure and high complexity of these documents. Previous approaches in this direction have relied on huge labelled datasets, using hand engineered features, leveraging on domain knowledge and focussed their attention on a narrow sub-domain for increased effectiveness. In this paper, we propose simple generic techniques using neural network for the summarization task for Indian legal judgment documents. We explore two neural network architectures for this task utilizing the word and sentence embeddings for capturing the semantics. The main advantage of the proposed approaches is that they do not rely on hand crafted features, or domain specific knowledge, nor is their application restricted to a particular sub-domain thus making them suitable to be extended to other domains as well. We tackle the problem of unavailability of labelled data for the task by assigning classes/scores to sentences in the training set, based on their match with reference summary produced by humans. The experimental evaluations establish the effectiveness of our proposed approaches as compared with other baselines.

Document clustering has recently been paid great attention in retrieval, navigation, and summarization of huge volumes of documents. With a better document clustering approach, computers can organize a document corpus automatically to a meaningful cluster for enabling efficient navigation, and browsing of the corpus. Document navigation and browsing is a valuable complement to the deficiencies of information retrieval technologies. This paper introduces Modsup-based frequent itemset and Rider Optimization-based Moth Search Algorithm (Rn-MSA) for clustering the documents. At first, the input documents are given to the pre-processing step, and then, the extraction is carried out based on TF-IDF and Wordnet features. Once the extraction is done, the feature selection is carried out based on frequent itemset for the establishment of feature knowledge. At last, the document clustering is done using the proposed Rn-MSA, which is designed by combining Rider Optimization Algorithm (ROA), and the Moth Search Algorithm (MSA). The performance of the document clustering based on proposed Modsup + Rn-MSA is evaluated in terms of precision, recall, F-Measure, and accuracy. The developed document clustering method achieves the maximal precision of 95.90%, maximal recall of 96.41%, maximal F-Measure of 96.41%, and the maximal accuracy of 95.12% that indicates its superiority.

Although video summarization has achieved tremendous success benefiting from Recurrent Neural Networks (RNN), RNN-based methods neglect the global dependencies and multi-hop relationships among video frames, which limits the performance. Transformer is an effective model to deal with this problem, and surpasses RNN-based methods in several sequence modeling tasks, such as machine translation, video captioning, etc. Motivated by the great success of transformer and the natural structure of video (frame-shot-video), a hierarchical transformer is developed for video summarization, which can capture the dependencies among frame and shots, and summarize the video by exploiting the scene information formed by shots. Furthermore, we argue that both the audio and visual information are essential for the video summarization task. To integrate the two kinds of information, they are encoded in a two-stream scheme, and a multimodal fusion mechanism is developed based on the hierarchical transformer. In this paper, the proposed method is denoted as Hierarchical Multimodal Transformer (HMT). Practically, extensive experiments show that HMT achieves (F-measure: 0.441, Kendall’s τ : 0.079, Spearman’s ρ : 0.080) and (F-measure: 0.601, Kendall’s τ : 0.096, Spearman’s ρ : 0.107) on SumMe and TVsum, respectively. It surpasses most of the traditional, RNN-based and attention-based video summarization methods.

Code summarization (aka comment generation) provides a high-level natural language description of the function performed by code, which can benefit the software maintenance, code categorization and retrieval. To the best of our knowledge, the state-of-the-art approaches follow an encoder-decoder framework which encodes source code into a hidden space and later decodes it into a natural language space. Such approaches suffer from the following drawbacks: (a) they are mainly input by representing code as a sequence of tokens while ignoring code hierarchy; (b) most of the encoders only input simple features (e.g., tokens) while ignoring the features that can help capture the correlations between comments and code; (c) the decoders are typically trained to predict subsequent words by maximizing the likelihood of subsequent ground truth words, while in real world, they are excepted to generate the entire word sequence from scratch. As a result, such drawbacks lead to inferior and inconsistent comment generation accuracy. To address the above limitations, this paper presents a new code summarization approach using hierarchical attention network by incorporating multiple code features, including type-augmented abstract syntax trees and program control flows. Such features, along with plain code sequences, are injected into a deep reinforcement learning (DRL) framework (e.g., actor-critic network) for comment generation. Our approach assigns weights (pays “attention”) to tokens and statements when constructing the code representation to reflect the hierarchical code structure under different contexts regarding code features (e.g., control flows and abstract syntax trees). Our reinforcement learning mechanism further strengthens the prediction results through the actor network and the critic network, where the actor network provides the confidence of predicting subsequent words based on the current state, and the critic network computes the reward values of all the possible extensions of the current state to provide global guidance for explorations. Eventually, we employ an advantage reward to train both networks and conduct a set of experiments on a real-world dataset. The experimental results demonstrate that our approach outperforms the baselines by around 22% to 45% in BLEU-1 and outperforms the state-of-the-art approaches by around 5% to 60% in terms of S-BLEU and C-BLEU.

Semantic segmentation is a critical component in autonomous driving and has to be thoroughly evaluated due to safety concerns. Deep neural network (DNN) based semantic segmentation models are widely used in autonomous driving. However, it is challenging to evaluate DNN-based models due to their black-box-like nature, and it is even more difficult to assess model performance for crucial objects, such as lost cargos and pedestrians, in autonomous driving applications. In this work, we propose VASS, a Visual Analytics approach to diagnosing and improving the accuracy and robustness of Semantic Segmentation models, especially for critical objects moving in various driving scenes. The key component of our approach is a context-aware spatial representation learning that extracts important spatial information of objects, such as position, size, and aspect ratio, with respect to given scene contexts. Based on this spatial representation, we first use it to create visual summarization to analyze models' performance. We then use it to guide the generation of adversarial examples to evaluate models' spatial robustness and obtain actionable insights. We demonstrate the effectiveness of VASS via two case studies of lost cargo detection and pedestrian detection in autonomous driving. For both cases, we show quantitative evaluation on the improvement of models' performance with actionable insights obtained from VASS.

Building a visual overview of temporal event sequences with an optimal level-of-detail (i.e. simplified but informative) is an ongoing challenge - expecting the user to zoom into every important aspect of the overview can lead to missing insights. We propose a technique to build a multilevel overview of event sequences, whose granularity can be transformed across sequence clusters (vertical level-of-detail) or longitudinally (horizontal level-of-detail), using hierarchical aggregation and a novel cluster data representation Align-Score-Simplify. By default, the overview shows an optimal number of sequence clusters obtained through the average silhouette width metric - then users are able to explore alternative optimal sequence clusterings. The vertical level-of-detail of the overview changes along with the number of clusters, whilst the horizontal level-of-detail refers to the level of summarization applied to each cluster representation. The proposed technique has been implemented into a visualization system called Sequence Cluster Explorer (Sequen-C) that allows multilevel and detail-on-demand exploration through three coordinated views, and the inspection of data attributes at cluster, unique sequence, and individual sequence level. We present two case studies using real-world datasets in the healthcare domain: CUREd and MIMIC-III; which demonstrate how the technique can aid users to obtain a summary of common and deviating pathways, and explore data attributes for selected patterns., Comment: This is the author's version of the article to be published in IEEE Transactions on Visualization and Computer Graphics

Due to recent advances in the film industry, the production of movies has grown exponentially, which has led to challenges in what is referred to as discoverability: given the overwhelming number of choices, choosing which film to watch has become a tedious task for audiences. Movie summarization (MS) could help, as it presents the central theme of the movie in a compact format and makes browsing more efficient for the audience. In this paper, we present an automatic MS framework coined as ‘QuickLook’, which identifies the leading characters and fuses multiple cues extracted from a movie. Firstly, the movie data is preprocessed for its division into scenes, followed by shot segmentation. Secondly, the leading characters in each segmented scene are determined. Next, four visual cues that capture the film's scenic beauty, memorability, informativeness and emotional resonance are extracted from shots containing the leading characters. These extracted features are then intelligently fused based on the assignment of different weights; shots with a fusion score above a certain threshold are selected for the final summary. The proposed MS framework is assessed by comparison with official trailers from ten Hollywood movies, providing a novel baseline for future fair comparison in the MS literature. The proposed framework is shown to outperform other state-of-the-art MS methods in terms of enjoyability and informativeness.

Nowadays, most research conducted in the field of abstractive text summarization focuses on neural-based models alone, without considering their combination with knowledge-based approaches that could further enhance their efficiency. In this direction, this work presents a novel framework that combines sequence-to-sequence neural-based text summarization along with structure and semantic-based methodologies. The proposed framework is capable of dealing with the problem of out-of-vocabulary or rare words, improving the performance of the deep learning models. The overall methodology is based on a well-defined theoretical model of knowledge-based content generalization and deep learning predictions for generating abstractive summaries. The framework is composed of three key elements: (i) a pre-processing task, (ii) a machine learning methodology, and (iii) a post-processing task. The pre-processing task is a knowledge-based approach, based on ontological knowledge resources, word sense disambiguation, and named entity recognition, along with content generalization, that transforms ordinary text into a generalized form. A deep learning model of attentive encoder-decoder architecture, which is expanded to enable a coping and coverage mechanism, as well as reinforcement learning and transformer-based architectures, is trained on a generalized version of text-summary pairs, learning to predict summaries in a generalized form. The post-processing task utilizes knowledge resources, word embeddings, word sense disambiguation, and heuristic algorithms based on text similarity methods in order to transform the generalized version of a predicted summary to a final, human-readable form. An extensive experimental procedure on three popular data sets evaluates key aspects of the proposed framework, while the obtained results exhibit promising performance, validating the robustness of the proposed approach.

The standard approach for abstractive text summarization is to use an encoder-decoder architecture. The encoder is responsible for capturing the general meaning from the source text, and the decoder is in charge of generating the final text summary. While this approach can compose summaries that resemble human writing, some may contain unrelated or unfaithful information. This problem is called “hallucination” and it represents a serious issue in legal texts as legal practitioners rely on these summaries when looking for precedents, used to support legal arguments. Another concern is that legal documents tend to be very long and may not be fed entirely to the encoder. We propose our method called LegalSumm for addressing these issues by creating different “views” over the source text, training summarization models to generate independent versions of summaries, and applying entailment module to judge how faithful these candidate summaries are with respect to the source text. We show that the proposed approach can select candidate summaries that improve ROUGE scores in all metrics evaluated.

More and more weakly structured, and irregular data sources are becoming available every day. The schema of these sources is useful for a number of tasks, such as query answering, exploration and summarization. However, although semantic web data might contain schema information, in many cases this is completely missing or partially defined. In this paper, we present a survey of the state of the art on schema information extraction approaches. We analyze and classify these approaches into three families: (1) approaches that exploit the implicit structure of the data, without assuming that some explicit statements on the schema are provided in the dataset; (2) approaches that use the explicit schema statements contained in the dataset to complement and enrich the schema, and (3) those that discover structural patterns contained in a dataset. We compare these studies in terms of their approach, advantages and limitations. Finally we discuss the problems that remain open.

It is essential to provide a structured data feed to the computer to accomplish any task so that it can process flawlessly to generate the desired output within minimal computational time. Generally, computer programmers should provide a structured data feed to the computer program for its successful execution. The hardcopy document should be scanned to generate its corresponding computer-readable softcopy version of the file. This process also proves to be a budget-friendly approach to disengage human resources from the entire process of record maintenance. Due to this automation, the workload of existing manpower is reduced to a significant level. This concept may prove beneficial for the delivery of any type of services to the ultimate beneficiary (i.e., citizen) in a minimal time frame. The administration has to deal with various issues of citizens due to the pressure of a huge population who seek legal help to resolve their issues, thereby leading to the filing of large numbers of pending legal cases at several courts of the country. To assist the victims with prompt delivery of justice and legal professionals in reducing their workload, this paper proposed a machine learning based automated legal model to enhance the efficiency of the legal support system with an accuracy of 94%.

Text summarization aims to reduce text by removing less useful information to obtain information quickly and precisely. In Indonesian abstractive text summarization, the research mostly focuses on multi-document summarization which methods will not work optimally in single-document summarization. As the public summarization datasets and works in English are focusing on single-document summarization, this study emphasized on Indonesian single-document summarization. Abstractive text summarization studies in English frequently use Bidirectional Encoder Representations from Transformers (BERT), and since Indonesian BERT checkpoint is available, it was employed in this study. This study investigated the use of Indonesian BERT in abstractive text summarization on the IndoSum dataset using the BERTSum model. The investigation proceeded by using various combinations of model encoders, model embedding sizes, and model decoders. Evaluation results showed that models with more embedding size and used Generative Pre-Training (GPT)-like decoder could improve the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) score and BERTScore of the model results.

Nowadays, opinion texts are quickly published on websites and social networks by various users in the form of short texts and also in high volumes and various fields. Because these texts reflect the opinions of many users, their processing and analysis, such as clustering, can be very useful in a variety of applications including politics, industry, commerce, and economics. High dimensions of the text representation decrease efficiency of clustering, and an effective solution for this challenge is reducing dimensions of texts. Manifold learning is a powerful tool for nonlinear dimension reduction of high-dimensional data. Therefore, in this paper, for increasing efficiency of opinion texts clustering, by manifold learning, dimensions of the represented opinion texts are reduced based on sentiment and semantics, and their intrinsic dimensions are extracted. Then, the clustering algorithm is applied to dimension-reduced opinion texts. The proposed approach helps us to cluster opinion texts with simultaneous consideration of sentiment and semantics, which has received very little attention in the previous works. This type of clustering helps users of opinion texts to obtain more useful information from texts and also provides more accurate summaries in applications, such as the summarization of opinion texts. Experimental results on three datasets show better performance of the proposed approach on opinion texts in terms of important measures for evaluating clustering efficiency. An improvement of about 9% is observed in terms of accuracy on the third dataset and clustering based on sentiment and semantics.

Neural sequence-to-sequence models are the state-of-the-art approach used in abstractive summarization of textual documents, useful for producing condensed versions of source text narratives without being restricted to using only words from the original text. Despite the advances in abstractive summarization, custom generation of summaries (e.g., towards a user’s preference) remains unexplored. In this article, we present CATS, an abstractive neural summarization model that summarizes content in a sequence-to-sequence fashion while also introducing a new mechanism to control the underlying latent topic distribution of the produced summaries. We empirically illustrate the efficacy of our model in producing customized summaries and present findings that facilitate the design of such systems. We use the well-known CNN/DailyMail dataset to evaluate our model. Furthermore, we present a transfer-learning method and demonstrate the effectiveness of our approach in a low resource setting, i.e., abstractive summarization of meetings minutes, where combining the main available meetings’ transcripts datasets, AMI and International Computer Science Institute(ICSI) , results in merely a few hundred training documents.

The traditional frequency based approach to creating multi-document extractive summary ranks sentences based on scores computed by summing up TF*IDF weights of words contained in the sentences. In this approach, TF or term frequency is calculated based on how frequently a term (word) occurs in the input and TF calculated in this way does not take into account the semantic relations among terms. In this paper, we propose methods that exploits semantic term relations for improving sentence ranking and redundancy removal steps of a summarization system. Our proposed summarization system has been tested on DUC 2003 and DUC 2004 benchmark multi-document summarization datasets. The experimental results reveal that performance of our multi-document text summarizer is significantly improved when the distributional term similarity measure is used for finding semantic term relations. Our multi-document text summarizer also outperforms some well known summarization baselines to which it is compared.

Headline generation is a challenging subtask of abstractive text summarization, which its output should be a summary, shorter than one sentence. It would be precious to develop a dataset for the evaluation of abstractive summarization methods on this task in the Persian language. There are several datasets for headline generation in Persian, most of which are not large enough to be used by more sophisticated methods of text summarization, such as deep learning models. Moreover, all of these datasets are focused on daily news and there is no dataset for summarizing scientific Persian papers. In this article, we present “ElmNet,” a headline generation dataset of about 400,000 abstract/headline pairs of scientific papers, gathered from six major publishers for scientific articles in Persian. We, moreover, evaluate the performance of the most important deep learning-based headline generation methods, on the proposed dataset. The results prove the comparability of the performance of the state-of-the-art methods on this task, to their results on the existing English datasets.

Features in a text are hierarchically structured and may not be optimally learned using one-step encoding. Scrutinizing the literature several times facilitates a better understanding of content and helps frame faithful context representations. The proposed model encapsulates the idea of re-examining a piece of text multiple times to grasp the underlying theme and aspects of English grammar before formulating a summary. We suggest a multi-level shared-weight encoder (MSE) that exclusively focuses on the sentence summarization task. MSE exercises a weight-sharing mechanism for proficiently regulating the multi-level encoding process. Weight-sharing helps recognize patterns left undiscovered by single level encoding strategy. We perform experiments with six encoding levels with weight sharing on the renowned short sentence summarization Gigaword and DUC2004 Task1 datasets. The experiments show that MSE generates a more readable(fluent) summary (Rouge-L score) as compared to multiple benchmark models while preserving similar levels of informativeness (Rouge-1 and Rouge-2 scores). Moreover, human evaluation of the generated abstracts also corroborates these assertions of enhanced readability.

Existing text summarization methods mainly rely on the mapping between manually labeled standard summaries and the original text for feature extraction, often ignoring the internal structure and semantic feature information of the original document. Therefore, the text summary extracted by the existing model has the problems of grammatical structure errors and semantic deviation from the original text. This paper attempts to enhance the model’s attention to the inherent feature information of the source text so that the model can more accurately identify the grammatical structure and semantic information of the document. Therefore, this paper proposes a model based on the multi-head self-attention mechanism and the soft attention mechanism. By introducing an improved multi-head self-attention mechanism in the model coding stage, the training model enables the correct summary syntax and semantic information to obtain higher weight, thereby making the generated summary more coherent and accurate. At the same time, the pointer network model is adopted, and the coverage mechanism is improved to solve out-of-vocabulary and repetitive problems when generating abstracts. This article uses CNN/DailyMail dataset to verify the model proposed in this article and uses the ROUGE indicator to evaluate the model. The experimental results show that the model in this article improves the quality of the generated summary compared with other models.

Multi-modal abstractive summarization for videos is an emerging task, aiming to integrate multi-modal and multi-source inputs (video, audio transcript) into a compressed textual summary. Although recent multi-encoder-decoder models on this task have shown promising performance, they did not explicitly model interactions of multi-source inputs. While some strategies like co-attention are utilized for modeling this interaction, considering ultra-long sequences and additional decoder in this task, the coupling of multi-modal data from multi-encoders and decoder needs complicated structure and additional parameters. In this paper, we propose a concise Decoder-only Multi-modal Transformer (D-MmT) based on the above observations. Specifically, we cut the encoder structure, and introduce an in-out shared multi-modal decoder to make the multi-source and target fully interact and couple in the shared feature space, reducing the model parameter redundancy. Also, we design a concise cascaded cross-modal interaction (CXMI) module in the multi-modal decoder that generates joint fusion representations and spontaneously establishes a fine-grained intra- and inter- association between multi-modalities. In addition, to make full use of the ultra-long sequence information, we introduce a joint in-out loss to make the input transcript also participate in backpropagation to enhance the contextual feature representation. The experimental results on the How2 dataset show that the proposed model outperforms the current state-of-the-art approach with fewer model parameters. Further analysis and visualization show the effectiveness of our proposed framework.

Application Programming Interfaces (APIs) have been widely discussed on social-technical platforms (e.g., Stack Overflow). Extracting API mentions from such informal software texts is the prerequisite for API-centric search and summarization of programming knowledge. Machine learning based API extraction has demonstrated superior performance than rule-based methods in informal software texts that lack consistent writing forms and annotations. However, machine learning based methods have a significant overhead in preparing training data and effective features. In this paper, we propose a multi-layer neural network based architecture for API extraction. Our architecture automatically learns character-, word- and sentence-level features from the input texts, thus removing the need for manual feature engineering and the dependence on advanced features (e.g., API gazetteers) beyond the input texts. We also propose to adopt transfer learning to adapt a source-library-trained model to a target-library, thus reducing the overhead of manual training-data labeling when the software text of multiple programming languages and libraries need to be processed. We conduct extensive experiments with six libraries of four programming languages which support diverse functionalities and have different API-naming and API-mention characteristics. Our experiments investigate the performance of our neural architecture for API extraction in informal software texts, the importance of different features, the effectiveness of transfer learning. Our results confirm not only the superior performance of our neural architecture than existing machine learning based methods for API extraction in informal software texts, but also the easy-to-deploy characteristic of our neural architecture.

In the era of big data, information overload problems are becoming increasingly prominent. It is challengingfor machines to understand, compress and filter massive text information through the use of artificial intelligencetechnology. The emergence of automatic text summarization mainly aims at solving the problem ofinformation overload, and it can be divided into two types: extractive and abstractive. The former finds somekey sentences or phrases from the original text and combines them into a summarization; the latter needs acomputer to understand the content of the original text and then uses the readable language for the human tosummarize the key information of the original text. This paper presents a two-stage optimization method forautomatic text summarization that combines abstractive summarization and extractive summarization. First,a sequence-to-sequence model with the attention mechanism is trained as a baseline model to generate initialsummarization. Second, it is updated and optimized directly on the ROUGE metric by using deep reinforcementlearning (DRL). Experimental results show that compared with the baseline model, Rouge-1, Rouge-2,and Rouge-L have been increased on the LCSTS dataset and CNN/DailyMail dataset.

In recent years, a large amount of data has been accumulated, such as those recorded in geological journals and report literature, which contain a wealth of information, but these data have not been fully exploited or mined. Automatic information extraction offers an effective way to achieve new discoveries and pursue further analysis, which is of great significance for users, researchers or decision makers to aid and support analysis. In this paper, we utilize the bidirectional encoder representations from transformers (BERT) model, which is fine-tuned and then applied to automatically generate the title of a given input summarization based on the collection of published literature samples. The framework contains an encoder module, decoder module and training module. The core stages of summary generation involve the combination of encoder and decoder modules, and the multi-stage function is then used to connect modules, thus endowing the text summarization model with a multi-task learning architecture. Compared to other baseline models, our proposed model obtains the best results on the constructed dataset. Therefore, based on the proposed model, an automatic geological briefing generation platform is developed and used as an online platform to support the excavation of key areas and a visual presentation analysis of the literature.

Creating a coherent summary of the text is a challenging task in the field of Natural Language Processing (NLP). Various Automatic Text Summarization techniques have been developed for abstractive as well as extractive summarization. This study focuses on extractive summarization which is a process containing selected delineative paragraphs or sentences from the original text and combining these into smaller forms than the document(s) to generate a summary. The methods that have been used for extractive summarization are based on a graph-theoretic approach, machine learning, Latent Semantic Analysis (LSA), neural networks, cluster, and fuzzy logic. In this paper, a semantic graph-based approach SGATS (Semantic Graph-based approach for Automatic Text Summarization) is proposed to generate an extractive summary. The proposed approach constructs a semantic graph of the original Hindi text document by establishing a semantic relationship between sentences of the document using Hindi Wordnet ontology as a background knowledge source. Once the semantic graph is constructed, fourteen different graph theoretical measures are applied to rank the document sentences depending on their semantic scores. The proposed approach is applied to two data sets of different domains of Tourism and Health. The performance of the proposed approach is compared with the state-of-the-art TextRank algorithm and human-annotated summary. The performance of the proposed system is evaluated using widely accepted ROUGE measures. The outcomes exhibit that our proposed system produces better results than TextRank for health domain corpus and comparable results for tourism corpus. Further, correlation coefficient methods are applied to find a correlation between eight different graphical measures and it is observed that most of the graphical measures are highly correlated.

Automatic text summarization extracts important information from texts and presents the information in the form of a summary. Abstractive summarization approaches progressed significantly by switching to deep neural networks, but results are not yet satisfactory, especially for languages where large training sets do not exist. In several natural language processing tasks, a cross-lingual model transfer is successfully applied in less-resource languages. For summarization, the cross-lingual model transfer was not attempted due to a non-reusable decoder side of neural models that cannot correct target language generation. In our work, we use a pre-trained English summarization model based on deep neural networks and sequence-to-sequence architecture to summarize Slovene news articles. We address the problem of inadequate decoder by using an additional language model for the evaluation of the generated text in target language. We test several cross-lingual summarization models with different amounts of target data for fine-tuning. We assess the models with automatic evaluation measures and conduct a small-scale human evaluation. Automatic evaluation shows that the summaries of our best cross-lingual model are useful and of quality similar to the model trained only in the target language. Human evaluation shows that our best model generates summaries with high accuracy and acceptable readability. However, similar to other abstractive models, our models are not perfect and may occasionally produce misleading or absurd content.

The learners and teachers of the teaching-learning process highly depend on online learning systems such as E-learning, which contains huge volumes of electronic contents related to a course. The multi-document summarization (MDS) is useful for summarizing such electronic contents. This article applies the task of MDS in an E-learning context. The objective of this article is threefold: 1) design a generic graph based multi-document summarizer DSGA (Dynamic Summary Generation Algorithm) to produce a variable length (dynamic) summary of academic text based learning materials based on a learner's request; 2) analyze the summary generation process; 3) perform content-based and task-based evaluations on the generated summary. The experimental results show that the DSGA summarizer performs better than the graph-based summarizers LexRank (LR) and Aggregate Similarity (AS). From the task-based evaluation, it is observed that the generated summary helps the learners to understand and comprehend the materials easily.

The automatic correction of grammar and spelling errors is important for students, second language learners, and some Natural Language Processing (NLP) tasks such as part of speech and text summarization. Recently, Neural Machine Translation (NMT) has been an out-performing and well-established model in the task of Grammar Error Correction (GEC). Arabic GEC is still growing because of some challenges, such as scarcity of training sets and the complexity of Arabic language. To overcome these issues, we introduced an unsupervised method to generate large-scale synthetic training data based on confusion function to increase the amount of training set. Furthermore, we introduced a supervised NMT model for AGEC called SCUT AGEC. SCUT AGEC is a convolutional sequence-to-sequence model consisting of nine encoder-decoder layers with attention mechanism. We applied fine-tuning to improve the performance and get more efficient results. Convolutional Neural Networks (CNN) gives our model ability to joint feature extraction and classification in one task and we proved that it is an efficient way to capture features of the local context. Moreover, it is easy to obtain long-term dependencies because of convolutional layers staking. Our proposed model becomes the first supervised AGEC system based on the convolutional sequence-to-sequence learning to outperforms the current state-of-the-art neural AGEC models.

Due to the massive amount of online textual data generated in a diversity of social media, web, and other information-centric applications. To select the vital data from the large text, need to study the full article and generate summary also not loose critical information of text document this process is called summarization. Text summarization is done either by human which need expertise in that area, also very tedious and time consuming. second type of summarization is done through system which is known as automatic text summarization which generate summary automatically. There are mainly two categories of Automatic text summarizations that is abstractive and extractive text summarization. Extractive summary is produced by picking important and high rank sentences and word from the text document on the other hand the sentences and word are present in the summary generated through Abstractive method may not present in original text. This article mainly focuses on different ATS (Automatic text summarization) techniques that has been instigated in the present are argue. The paper begin with a concise introduction of automatic text summarization, then closely discussed the innovative developments in extractive and abstractive text summarization methods, and then transfers to literature survey, and it finally sum-up with the proposed techniques using LSTM with encoder Decoder for abstractive text summarization are discussed along with some future work directions.

Emperor Penguin Optimizer (EPO) is a recently developed metaheuristic algorithm to solve general optimization problems. The main strength of EPO is twofold. Firstly, EPO has low learning curve (i.e., based on the simple analogy of huddling behavior of emperor penguins in nature (i.e., surviving strategy during Antarctic winter). Secondly, EPO offers straightforward implementation. In the EPO, the emperor penguins represent the candidate solution, huddle denotes the search space that comprises a two-dimensional L-shape polygon plane, and randomly positioned of the emperor penguins represents the feasible solution. Among all the emperor penguins, the focus is to locate an effective mover representing the global optimal solution. To-date, EPO has slowly gaining considerable momentum owing to its successful adoption in many broad range of optimization problems, that is, from medical data classification, economic load dispatch problem, engineering design problems, face recognition, multilevel thresholding for color image segmentation, high-dimensional biomedical data analysis for microarray cancer classification, automatic feature selection, event recognition and summarization, smart grid system, and traffic management system to name a few. Reflecting on recent progress, this paper thoroughly presents an in-depth study related to the current EPO’s adoption in the scientific literature. In addition to highlighting new potential areas for improvements (and omission), the finding of this study can serve as guidelines for researchers and practitioners to improve the current state-of-the-arts and state-of-practices on general adoption of EPO while highlighting its new emerging areas of applications.

Bio-loggers are widely used for studying the movement and behavior of animals. However, some sensors provide more data than is practical to store given experiment or bio-logger design constraints. One approach for overcoming this limitation is to utilize data collection strategies, such as non-continuous recording or data summarization that may record data more efficiently, but need to be validated for correctness. In this paper we address two fundamental questions—how can researchers determine suitable parameters and behaviors for bio-logger sensors, and how do they validate their choices? We present a methodology that uses software-based simulation of bio-loggers to validate various data collection strategies using recorded data and synchronized, annotated video. The use of simulation allows for fast and repeatable tests, which facilitates the validation of data collection methods as well as the configuration of bio-loggers in preparation for experiments. We demonstrate this methodology using accelerometer loggers for recording the activity of the small songbird Junco hyemalis hyemalis.

Determining the extent to which two text snippets are semantically equivalent is a well-researched topic in the areas of natural language processing, information retrieval and text summarization. The sentence-to-sentence similarity scoring is extensively used in both generic and query-based summarization of documents as a significance or a similarity indicator. Nevertheless, most of these applications utilize the concept of semantic similarity measure only as a tool, without paying importance to the inherent properties of such tools that ultimately restrict the scope and technical soundness of the underlined applications. This paper aims to contribute to fill in this gap. It investigates three popular WordNet hierarchical semantic similarity measures, namely path-length, Wu and Palmer and Leacock and Chodorow, from both algebraical and intuitive properties, highlighting their inherent limitations and theoretical constraints. We have especially examined properties related to range and scope of the semantic similarity score, incremental monotonicity evolution, monotonicity with respect to hyponymy/hypernymy relationship as well as a set of interactive properties. Extension from word semantic similarity to sentence similarity has also been investigated using a pairwise canonical extension. Properties of the underlined sentence-to-sentence similarity are examined and scrutinized. Next, to overcome inherent limitations of WordNet semantic similarity in terms of accounting for various Part-of-Speech word categories, a WordNet “All word-To-Noun conversion” that makes use of Categorial Variation Database (CatVar) is put forward and evaluated using a publicly available dataset with a comparison with some state-of-the-art methods. The finding demonstrates the feasibility of the proposal and opens up new opportunities in information retrieval and natural language processing tasks.

Text summarization is one problem in natural language processing that generates a brief version of the original document. This research took attention for some researchers in this last decade and growing fast, including Indonesia language. This paper aims to recap summarization text research especially in Indonesia language. As usual, this paper discusses two summarization approaches, extractive and abstractive. In fact, the number of research of extractive is more than abstractive. This paper investigates some methods such as Statistical Based Approach, Graph Based Approach, Machine Learning Approach, Fuzzy Logic Approach, Algebraic Approach, and Hybrid Approach. This paper shows some methods details and summarize the results. Keywords— Text summarization, extractive summary, abstractive summary, natural language processing

The Seq2Seq abstract summarization model based on long short-term memory (LSTM) is very effective for short text summarization. However, LSTM is limited by long-term dependencies, which can potentially result in salient information loss when long text is processed by the Seq2Seq model based on LSTM. To overcome the long-term dependence limitation, an encoder-decoder model based on the dynamic residual network is proposed in this work. The model can dynamically select an optimal state from the state history to establish a connection with the current state to improve the LSTM long sequence dependencies according to the current decoding environment. Because the dynamic residual connections will result in long-term connection-dependent words, a new method based on reinforcement learning is proposed to simulate the dependence between words, which is then implemented into the training process of the model. This model is verified using the CNN/Daily Mail and New York Times datasets, and the experimental results show that the proposed model achieves significant improvements in capturing long-term dependencies compared with the traditional LSTM-based Seq2Seq abstractive summarization model.

Summary writing is essential for academic success, and has attracted renewed interest in academic research and large-scale language test. However, less attention has been paid to the development and evaluation of the scoring scales of summary writing. This study reports on the validation of a summary rubric that represented an approach to scale development with limited resources out of consideration for practicality. Participants were 83 students and three raters. Diagnostic evaluation of the scale components and categories was based on raters’ perception of their use and the scores of students’ summaries which were analyzed using multifaceted Rasch measurement (MFRM). Correlation analysis revealed significant relationships among the scoring components, but the coefficients among some of the components were over high. MFRM analysis provided evidence in support of the usefulness of the scoring rubric, but also suggested the need of a refinement of the components and categories. According to the raters, the rubric was ambiguous in addressing some crucial text features. This study has implications for summarization task design, scoring scale development and validation in particular.

Rapid and exponential development of textual data in recent years has yielded to the need for automatic text summarization models which aim to automatically condense a piece of text into a shorter version. Although various unsupervised and machine learning-based approaches have been introduced for text summarization during the last decades, the emergence of deep learning has made remarkable progress in this field. However, deep learning-based text summarization models are still in their early steps of development and their potential has yet to be fully explored. Accordingly, a novel abstractive summarization model is proposed in this paper which utilized the combination of convolutional neural network and long short-term memory integrated with auxiliary attention in its encoder to increase the saliency and coherency of generated summaries. The proposed model was validated on CNN\Daily Mail and DUC-2004 datasets and empirical results indicated that not only the proposed model outperformed existing models in terms of ROUGE metric but also its generated summaries had higher saliency and readability compared to the baseline model according to human evaluation.

The exponential growth of online textual data triggered the crucial need for an effective and powerful tool that automatically provides the desired content in a summarized form while preserving core information. In this paper, we propose an automatic, generic, and extractive Arabic single document summarizing method aiming at producing a sufficiently informative summary. The proposed extractive method evaluates each sentence based on a combination of statistical and semantic features in which a novel formulation is used taking into account sentence importance, coverage and diversity. Further, two summarizing techniques including score-based and supervised machine learning were employed to produce the summary and then assist leveraging the designed features. We demonstrate the effectiveness of the proposed method through a set of experiments under EASC corpus using ROUGE measure. Compared to some existing related work, the experimental evaluation shows the strength of the proposed method in terms of precision, recall, and F-score performance metrics.