Text summarization aims to extract essential information from a piece of text and transform it into a concise version. Existing unsupervised abstractive summarization models use recurrent neural networks framework and ignore abundant unlabeled corpora resources. In order to address these issues, we propose TED, a transformer-based unsupervised summarization system with pretraining on large-scale data. We first leverage the lead bias in news articles to pretrain the model on large-scale corpora. Then, we finetune TED on target domains through theme modeling and a denoising autoencoder to enhance the quality of summaries. Notably, TED outperforms all unsupervised abstractive baselines on NYT, CNN/DM and English Gigaword datasets with various document styles. Further analysis shows that the summaries generated by TED are abstractive and containing even higher proportions of novel tokens than those from supervised models.
Online propaganda is central to the recruitment strategies of extremist groups and in recent years these efforts have increasingly extended to women. To investigate ISIS' approach to targeting women in their online propaganda and uncover implications for counterterrorism, we rely on text mining and natural language processing (NLP). Specifically, we extract articles published in Dabiq and Rumiyah (ISIS's online English language publications) to identify prominent topics. To identify similarities or differences between these texts and those produced by non-violent religious groups, we extend the analysis to articles from a Catholic forum dedicated to women. We also perform an emotional analysis of both of these resources to better understand the emotional components of propaganda. We rely on Depechemood (a lexical-base emotion analysis method) to detect emotions most likely to be evoked in readers of these materials. The findings indicate that the emotional appeal of ISIS and Catholic materials are similar
Pathology reports contain useful information such as the main involved organ, diagnosis, etc. These information can be identified from the free text reports and used for large-scale statistical analysis or serve as annotation for other modalities such as pathology slides images. However, manual classification for a huge number of reports on multiple tasks is labor-intensive. In this paper, we have developed an automatic text classifier based on BERT and we propose a human-centric metric to evaluate the model. According to the model confidence, we identify low-confidence cases that require further expert annotation and high-confidence cases that are automatically classified. We report the percentage of low-confidence cases and the performance of automatically classified cases. On the high-confidence cases, the model achieves classification accuracy comparable to pathologists. This leads a potential of reducing 80% to 98% of the manual annotation workload.
We study the problem of identity testing of markov chains. In this setting, we are given access to a single trajectory from a markov chain with unknown transition matrix $Q$ and the goal is to determine whether $Q = P$ for some known matrix $P$ or $\text{Dist}(P, Q) \geq \epsilon$ where $\text{Dist}$ is suitably defined. In recent work by Daskalakis, Dikkala and Gravin, 2018, it was shown that it is possible to distinguish between the two cases provided the length of the observed trajectory is at least super-linear in the hitting time of $P$ which may be arbitrarily large. In this paper, we propose an algorithm that avoids this dependence on hitting time thus enabling efficient testing of markov chains even in cases where it is infeasible to observe every state in the chain. Our algorithm is based on combining classical ideas from approximation algorithms with techniques for the spectral analysis of markov chains.
A word having multiple senses in a text introduces the lexical semantic task to find out which particular sense is appropriate for the given context. One such task is Word sense disambiguation which refers to the identification of the most appropriate meaning of the polysemous word in a given context using computational algorithms. The language processing research in Hindi, the official language of India, and other Indian languages is restricted by unavailability of the standard corpus. For Hindi word sense disambiguation also, the large corpus is not available. In this work, we prepared the text containing new senses of certain words leading to the enrichment of the sense-tagged Hindi corpus of sixty polysemous words. Furthermore, we analyzed two novel lexical associations for Hindi word sense disambiguation based on the contextual features of the polysemous word. The evaluation of these methods is carried out over learning algorithms and favorable results are achieved.
Opinion polls have been the bridge between public opinion and politicians in elections. However, developing surveys to disclose people's feedback with respect to economic issues is limited, expensive, and time-consuming. In recent years, social media such as Twitter has enabled people to share their opinions regarding elections. Social media has provided a platform for collecting a large amount of social media data. This paper proposes a computational public opinion mining approach to explore the discussion of economic issues in social media during an election. Current related studies use text mining methods independently for election analysis and election prediction; this research combines two text mining methods: sentiment analysis and topic modeling. The proposed approach has effectively been deployed on millions of tweets to analyze economic concerns of people during the 2012 US presidential election.
In order for people to be able to trust and take advantage of the results of advanced machine learning and artificial intelligence solutions for real decision making, people need to be able to understand the machine rationale for given output. Research in explain artificial intelligence (XAI) addresses the aim, but there is a need for evaluation of human relevance and understandability of explanations. Our work contributes a novel methodology for evaluating the quality or human interpretability of explanations for machine learning models. We present an evaluation benchmark for instance explanations from text and image classifiers. The explanation meta-data in this benchmark is generated from user annotations of image and text samples. We describe the benchmark and demonstrate its utility by a quantitative evaluation on explanations generated from a recent machine learning algorithm. This research demonstrates how human-grounded evaluation could be used as a measure to qualify local machine-learning explanations.
Object referring has important applications, especially for human-machine interaction. While having received great attention, the task is mainly attacked with written language (text) as input rather than spoken language (speech), which is more natural. This paper investigates Object Referring with Spoken Language (ORSpoken) by presenting two datasets and one novel approach. Objects are annotated with their locations in images, text descriptions and speech descriptions. This makes the datasets ideal for multi-modality learning. The approach is developed by carefully taking down ORSpoken problem into three sub-problems and introducing task-specific vision-language interactions at the corresponding levels. Experiments show that our method outperforms competing methods consistently and significantly. The approach is also evaluated in the presence of audio noise, showing the efficacy of the proposed vision-language interaction methods in counteracting background noise.
An increasing amount of digital music is being published daily. Music streaming services often ingest all available music, but this poses a challenge: how to recommend new artists for which prior knowledge is scarce? In this work we aim to address this so-called cold-start problem by combining text and audio information with user feedback data using deep network architectures. Our method is divided into three steps. First, artist embeddings are learned from biographies by combining semantics, text features, and aggregated usage data. Second, track embeddings are learned from the audio signal and available feedback data. Finally, artist and track embeddings are combined in a multimodal network. Results suggest that both splitting the recommendation problem between feature levels (i.e., artist metadata and audio track), and merging feature embeddings in a multimodal approach improve the accuracy of the recommendations.
Add to Chrome
Add to Firefox