News Corpus Research Articles

Automatic detection of satire in Twitter: A psycholinguistic-based approach

In recent years, a substantial effort has been made to develop sophisticated methods that can be used to detect figurative language, and more specifically, irony and sarcasm. There is, however, an absence of new approaches and research works that analyze satirical texts. The recognition of satire by sentiment analysis and Natural Language Processing (NLP) applications is extremely important because it can influence and change the meaning of a statement in varied and complex ways. We used this understanding as a basis to propose a method that employs a wide variety of psycholinguistic features and which detects satirical and non-satirical text. We then went on to train a set of machine learning algorithms that would enable us to classify unknown data. Finally, we conducted several experiments in order to detect the most relevant features that generate a better pattern as regards detecting satirical texts. We evaluated the effectiveness of our method by obtaining a corpus of satirical and non-satirical news from Mexican and Spanish Twitter accounts. Our proposal obtained encouraging results, with an F-measure of 85.5% for Mexico and one of 84.0% for Spain. Moreover, the results of the experiment showed that there is no significant difference between Mexican and Spanish satire.

Model of Network Topic Detection Based on Web Usage Behaviour Mode Analysis and Mining Technology

This research has caught researchers’ wide attention for detecting network topic exactly with the arrival of big data era characterized by semi-structured or unstructured text. This paper proposes a model of network topic detection based on web usage behaviour mode analysis and mining technology taking Web news as object of research. The author elaborates main function and method proposed in this model, which include the analysis module of Web news instance clicking mode, the analysis module of Web news instance retrieval mode, the analysis module of Web news instance seed and the analysis module of similar Web news instance supporting topics. Based on these functions and methods, the author elaborates main algorithm proposed in this model, which include the mining algorithm of Web news seed instances and the mining algorithm of similar Web news instances supporting topics. These functional algorithms have been applied in processing module of model, and focus on how to detect network topic efficiently from a large number of web usage behaviour towards to Web news instances, in order to explore a research method for network topic detection. The process of experimental analysis includes three steps, firstly, the author analyses the precision of topic detection under different method, secondly, the author completes the impact analysis of Web news topic detection quality from the number of Web news instances concerned and seed threshold, finally, the author completes the quality impact analysis of Web news instances mined supporting topic from the number of Web news instances concerned and probability threshold. The results of experimental analysis show the feasibility, validity and superiority of model design and play an important role in constructing topic-focused Web news corpus so as to provide a real-time data source for topic evolution tracking.

Variational Autoencoder for Semi-Supervised Text Classification

Although semi-supervised variational autoencoder (SemiVAE) works in image classification task, it fails in text classification task if using vanilla LSTM as its decoder. From a perspective of reinforcement learning, it is verified that the decoder's capability to distinguish between different categorical labels is essential. Therefore, Semi-supervised Sequential Variational Autoencoder (SSVAE) is proposed, which increases the capability by feeding label into its decoder RNN at each time-step. Two specific decoder structures are investigated and both of them are verified to be effective. Besides, in order to reduce the computational complexity in training, a novel optimization method is proposed, which estimates the gradient of the unlabeled objective function by sampling, along with two variance reduction techniques. Experimental results on Large Movie Review Dataset (IMDB) and AG's News corpus show that the proposed approach significantly improves the classification accuracy compared with pure-supervised classifiers, and achieves competitive performance against previous advanced methods. State-of-the-art results can be obtained by integrating other pretraining-based methods.

Keyword and Event Extraction for Thematic Map Retrieval from Indonesian Online News Site

Online news sites provide great deal of information that may be extracted and mined to create Geographical Information Retrieval (GIR) system that provides thematic map based on user query. The task requires extracting generic events and its attributes from the news corpus. This event extraction requires the keywords and topics that can help increase the accuracy of the extraction and retrieval process. We prepare a large online news corpus and compared the Latent Dirichlet Allocation (LDA) and CBOW and skip-gram model to help providing base thematic keywords which can assist the extraction process on the intended GIR system. LDA is better in terms of semantic relatedness and the CBOW and skip-gram is useful for providing semantic similarity.

Modeling Latent Topics and Temporal Distance for Story Segmentation of Broadcast News

This paper studies a strategy to model latent topics and temporal distance of text blocks for story segmentation, that we call graph regularization in topic modeling or GRTM. We propose two novel approaches that consider both temporal distance and lexical similarity of text blocks, collectively referred to as data proximity, in learning latent topic representation, where a graph regularizer is involved to derive the latent topic representation while preserving data proximity. In the first approach, we extend the idea of Laplacian probabilistic latent semantic analysis LapPLSA by introducing a distance penalty function in the affinity matrix of a graph for latent topic estimation. The estimated latent topic distributions are used to replace the traditional term-frequency vectors as the data representation of the text blocks and to measure the cohesive strength between them. In the second approach, we perform Laplacian eigenmaps, which makes use of the graph regularizer for dimensionality reduction, on latent topic distributions estimated by conventional topic modeling. We conduct the experiments on the automatic speech recognition transcripts of the TDT2 English broadcast news corpus. The experiments show the proposed strategy outperforms the conventional techniques. LapPLSA performs the best with the highest F1-measure of 0.816. The effects of the penalty constant in the distance penalty function, the number of latent topics, and the size of training data on the segmentation performances are also studied.

Company Name Discrimination in Tweets using Topic Signatures Extracted from News Corpus

It is impossible for any human being to analyze the more than 500 million tweets that are generated per day. Lexical ambiguities on Twitter make it difficult to retrieve the desired data and relevant topics. Most of the solutions for the word sense disambiguation problem rely on knowledge base systems. Unfortunately, it is expensive and time-consuming to manually create a knowledge base system, resulting in a knowledge acquisition bottleneck. To solve the knowledge-acquisition bottleneck, a topic signature is used to disambiguate words. In this paper, we evaluate the effectiveness of various features of newspapers on the topic signature extraction for word sense discrimination in tweets. Based on our results, topic signatures obtained from a snippet feature exhibit higher accuracy in discriminating company names than those from the article body. We conclude that topic signatures extracted from news articles improve the accuracy of word sense discrimination in the automated analysis of tweets.

Hierarchical representation and estimation of prosody using continuous wavelet transform

Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.

Study of sub-word acoustical models for Kannada isolated word recognition system

The speech recognition system basically extracts the textual information present in the speech. In the present work, speaker independent isolated word recognition system for one of the south Indian language--Kannada has been developed. For European languages such as English, large amount of research has been carried out in the context of speech recognition. But, speech recognition in Indian languages such as Kannada reported significantly less amount of work and there are no standard speech corpus readily available. In the present study, speech database has been developed by recording the speech utterances of regional Kannada news corpus of different speakers. The speech recognition system has been implemented using the Hidden Markov Tool Kit. Two separate pronunciation dictionaries namely phone based and syllable based dictionaries are built in-order to design and evaluate the performances of phone-level and syllable-level sub-word acoustical models. Experiments have been carried out and results are analyzed by varying the number of Gaussian mixtures in each state of monophone Hidden Markov Model (HMM). Also, context dependent triphone HMM models have been built for the same Kannada speech corpus and the recognition accuracies are comparatively analyzed. Mel frequency cepstral coefficients along with their first and second derivative coefficients are used as feature vectors and are computed in acoustic front-end processing. The overall word recognition accuracy of 60.2 and 74.35 % respectively for monophone and triphone models have been obtained. The study shows a good improvement in the accuracy of isolated-word Kannada speech recognition system using triphone HMM models compared to that of monophone HMM models.

Improved automatic English proficiency rating of unconstrained speech with multiple corpora

The performance of machine learning classifiers in automatically scoring the English proficiency of unconstrained speech has been explored. Suprasegmental measures were computed by software, which identifies the basic elements of Brazil's model in human discourse. This paper explores machine learning training with multiple corpora to improve two of those algorithms: prominent syllable detection and tone choice classification. The results show that machine learning training with the Boston University Radio News Corpus can improve automatic English proficiency scoring of unconstrained speech from a Pearson's correlation of 0.677---0.718. This correlation is higher than any other existing computer programs for automatically scoring the proficiency of unconstrained speech and is approaching that of human raters in terms of inter-rater reliability.

The coverage of science in television news programs in Brazil and Colombia: a comparative study of media constructs

This study analyzes and compares the science and technology coverage in Brazil's main television news program (Jornal Nacional) and its Colombian counterpart (Noticias Caracol). Using content analysis, we investigated a corpus of news stories broadcast from April 2009 to March 2010. We found that Jornal Nacional presented over twice as many reports on science and technology as Noticias Caracol, and that its levels of reporting remained fairly stable throughout the year. The Brazilian reports were also longer, were featured more prominently, and used more visual resources. Even so, some similarities were found: news about health and medicine was most frequent; the reports focused primarily on announcing new research; scientists were the main sources cited; and national research was prioritized.

Investigating evaluation and news values in news items that are shared through social media

The sharing of news through social media platforms is now a significant part of mainstream online media use and is an increasingly important consideration in journalism practice and production. This paper analyses the linguistic characteristics of online news sharing on Facebook, with a focus on evaluation and news values in a corpus of the 100 ‘most shared’ news items from ‘heritage’ English-language news media organisations. Analyses combine corpus linguistic techniques (semantic tagging, frequency analysis, concordancing) with manual, computer-aided annotation. The main focus is on discursive news values analysis (DNVA), which examines how news values are established through semiotic resources, enabling new empirical insights into shared news and adding a specific linguistic focus to the emerging literature on news sharing. Results suggest that all ‘traditional’ news values appear to be construed in the shared news corpus and that there is variety in terms of the items that are widely shared. At the same time, the news values of Eliteness, Superlativeness, Unexpectedness, Negativity and Timeliness seem especially important in the corpus. The findings also indicate that ‘unexpected’ and ‘affective’ news items may be shared more, and that Negativity is a more important news value than Positivity.

Análisis de las noticias sobre lectura en Portugal: Jornal de Notícias (2012)

This paper deals with the analysis of the 35 news about reading in the Portuguese newspaper Jornal de Notícias in 2012. The goal is to find out if new technologies are influencing the reading habits. The analysis is based on the study of the frequency rate and semantic value of a key words selection – present in the news corpus – for the object of the study. They allow observing the public discourses evolution along the period of time studied as a consequence of the new technologies development. The reading concept evolution along this period is also analyzed as well as the issues related to the reading concept such as reading sociology, its relationship with education and also the new reading formats.

Análisis de las noticias sobre lectura en Portugal: Jornal de Notícias (2012)

This paper deals with the analysis of the 35 news about reading in the Portuguese newspaper Jornal de Notícias in 2012. The goal is to find out if new technologies are influencing the reading habits. The analysis is based on the study of the frequency rate and semantic value of a key words selection – present in the news corpus – for the object of the study. They allow observing the public discourses evolution along the period of time studied as a consequence of the new technologies development. The reading concept evolution along this period is also analyzed as well as the issues related to the reading concept such as reading sociology, its relationship with education and also the new reading formats.

Corpus - based Translation of Korean Weather News into English

Weather news is reported several times every day to convey constantly changing weather conditions, which have an influence on our daily life. Thus, weather news has practical value to us. This study aims to translate diverse linguistic units that are used in Korean weather news into English on the basis of the corpus of Korean weather news and that of English weather news. Both corpora are constructed through the collection of online news from web sites, and thus they contain vivid expressions that are really used in the domain of weather news. This method of corpus-based translation enables us to obtain the product of translation that is free from translationese and to construct a parallel corpus that is useful to machine translation.

Arabic phonemes recognition using hybrid LVQ/HMM model for continuous speech recognition

In attempt to increase the rate of Arabic phonemes recognition, we introduce a novel hybrid recognition algorithm. The algorithm is composed of the learning vector quantization (LVQ) and hidden Markov model (HMM). The hybrid algorithm used to recognizing Arabic phonemes in continuous open-vocabulary speech. A recorded Arabic corpus of different TV news for modern standard Arabic was used for training and testing purposes. We employ a data driven approach to generate the training feature vectors that embed the frame neighboring correlation information. Next, we generate the phonemes codebooks using the K-means splitting algorithm. Then, we trained the generated codebooks using the LVQ algorithm. We achieved a performance of 98.49 % during independent classification training and 90 % during dependent classification training. When using the trained LVQ codebooks in Arabic utterance transcription, the phoneme recognition rate was 72 % using LVQ only. We combined the LVQ codebooks with the single state HMM model using enhanced Viterbi algorithm which includes the phonemes bigrams. We achieved 89 % of Arabic phonemes recognition rate based on the hybrid LVQ/HMM algorithm.

Leaders or readers, whom to please? News values in the transition of the Chinese press

This paper takes news values as the analytical point of entry to understand the working of Chinese press in the wake of its marketization reform. The reform has resulted Chinese newspapers in facing an increasingly tightening tension between the political power of the government and economic forces of the market. Drawing on four different but complementary datasets – news policy documents, media editorial meetings, interviews with Chinese journalism practitioners and a newly constructed corpus of Chinese news reporting, this study found that maneuvering news values concerning different news actors and events constituted a distinctive way for Chinese journalism practitioners to circumvent the dilemma. To please leaders, the news values of Eliteness and Positivity were foregrounded, while to please readers, the news values of Personalization and Negativity were highlighted. The findings are important for understanding Chinese journalists’ mindsets to accommodate structural changes of Chinese press in journalistic practices.

Merging open knowledge extracted from text with MERGILO

This paper presents MERGILO, a method for reconciling knowledge extracted from multiple natural language sources, and for delivering it as a knowledge graph. The underlying problem is relevant in many application scenarios requiring the creation and dynamic evolution of a knowledge base, e.g. automatic news summarization, human–robot dialoguing, etc. After providing a formal definition of the problem, we propose our holistic approach to handle natural language input – typically independent texts as in news from different sources – and we output a knowledge graph representing their reconciled knowledge. MERGILO is evaluated on its ability to identify corresponding entities and events across documents against a manually annotated corpus of news, showing promising results.

An Analysis of Adjectives Describing Meteorological Phenomena

Diverse meteorological phenomena such as rain, snow, sleet, wind, thunderstorm, hurricane, fog, and cloud are mentioned in daily weather news. This study aimed to analyze adjectives used for describing meteorological phenomena. To analyze these adjectives I first constructed a corpus of weather news obtained from web sites. Then I extracted adjectives describing meteorological phenomena from the corpus by following several steps that included the use of a concordancing software and a POS (part-of-speech) tagger. After that, on the basis of their meanings related to each meteorological phenomenon, these adjectives are classified into seven groups: Intensity, Time, Area, Temperature, Duration, Property, and Subjective Feeling. For instance, adjectives such as heavy, severe, thick, moderate, and light are classified into the group Intensity. Depending on meteorological phenomena, the range of the use of each adjective describing them appears to vary. That is, some adjectives are used to describe only a certain phenomenon, while others are employed to describe many. Therefore, this study systematically shows the types of adjectives used for describing diverse meteorological phenomena and restrictions on the use of them.

Multilingual news extraction via stopword language model scoring

Web news provides a quick and convenient means to create collections of large documents. The creation of a web news corpus has typically required the construction of a set of HTML parsing rules to identify content text. In general, these parsing rules are written manually and treat different web pages differently. We address this issue and propose a news content recognition algorithm that is language and layout independent. Our method first scans a given HTML document and roughly localizes a set of candidate news areas. Next, we apply a designed scoring function to rank the best content. To validate this approach, we evaluate the systems performance using 1092 items of multilingual web news data covering 17 global regions and 11 distinct languages. We compare these data with nine published content extraction systems using standard settings. The results of this empirical study show that our method outperforms the second-best approach (Boilerpipe) by 6.04 and 10.79 % with regard to the relative micro and macro F-measures, respectively. We also apply our system to monitor online RSS news distribution. It collected 0.4 million news articles from 200 RSS channels in 20 days. This sample quality test shows that our method achieved 93 % extraction accuracy for large news streams.

BenLem (A Bengali Lemmatizer) and Its Role in WSD

A lemmatization algorithm for Bengali has been developed and evaluated. Its effectiveness for word sense disambiguation (WSD) is also investigated. One of the key challenges for computer processing of highly inflected languages is to deal with the frequent morphological variations of the root words appearing in the text. Therefore, a lemmatizer is essential for developing natural language processing (NLP) tools for such languages. In this experiment, Bengali, which is the national language of Bangladesh and the second most popular language in the Indian subcontinent, has been taken as a reference. In order to design the Bengali lemmatizer (named as BenLem), possible transformations through which surface words are formed from lemmas are studied so that appropriate reverse transformations can be applied on a surface word to get the corresponding lemma back. BenLem is found to be capable of handling both inflectional and derivational morphology in Bengali. It is evaluated on a set of 18 news articles taken from the FIRE Bengali News Corpus consisting of 3,342 surface words (excluding proper nouns) and found to be 81.95% accurate. The role of the lemmatizer is then investigated for Bengali WSD. Ten highly polysemous Bengali words are considered for sense disambiguation. The FIRE corpus and a collection of Tagore’s short stories are considered for creating the WSD dataset. Different WSD systems are considered for this experiment, and it is noticed that BenLem improves the performance of all the WSD systems and the improvements are statistically significant.