Search

This paper proposes a novel method for generating compact answers to open-domain why-questions, such as the following answer, “Because deep learning technologies were introduced,” to the question, “Why did Google’s machine translation service improve so drastically?” Although many works have dealt with why-question answering, most have focused on retrieving as answers relatively long text passages that consist of several sentences. Because of their length, such passages are not appropriate to be read aloud by spoken dialog systems and smart speakers; hence, we need to create a method that generates compact answers. We developed a novel neural summarizer for this compact answer generation task. It combines a recurrent neural network-based encoderdecoder model with stacked convolutional neural networks and was designed to effectively exploit background knowledge, in this case a set of causal relations (e.g., “[Microsoft’s machine translation has made great progress over the last few years]effect since [it started to use deep learning.]cause”) that was extracted from a large web data archive (4 billion web pages). Our experimental results show that our method achieved significantly better ROUGE F-scores than existing encoder-decoder models and their variations that were augmented with query-attention and memory networks, which are used to exploit the background knowledge.

Extracting clinical terms from free-text format radiology reports is a first important step toward their secondary use. However, there is no general consensus on the kind of terms to be extracted. In this paper, we propose an information model comprising three types of clinical entities: observations, clinical findings, and modifiers. Furthermore, to determine its applicability for in-house radiology reports, we extracted clinical terms with state-of-the-art deep learning models and compared the results. We trained and evaluated models using 540 in-house chest computed tomography (CT) reports annotated by multiple medical experts. Two deep learning models were compared, and the effect of pre-training was explored. To investigate the generalizability of the model, we evaluated the use of other institutional chest CT reports. The micro F1-score of our best performance model using in-house and external datasets were 95.36% and 94.62%, respectively. Our results indicated that entities defined in our information model were suitable for extracting clinical terms from radiology reports, and the model was sufficiently generalizable to be used with dataset from other institutions.

In this paper, we propose a method for why-question answering (why-QA) that uses an adversarial learning framework. Existing why-QA methods retrieve “answer passages” that usually consist of several sentences. These multi-sentence passages contain not only the reason sought by a why-question and its connection to the why-question, but also redundant and/or unrelated parts. We use our proposed “Adversarial networks for Generating compact-answer Representation” (AGR) to generate from a passage a vector representation of the non-redundant reason sought by a why-question and exploit the representation for judging whether the passage actually answers the why-question. Through a series of experiments using Japanese why-QA datasets, we show that these representations improve the performance of our why-QA neural model as well as that of a BERT-based why-QA model. We show that they also improve a state-of-the-art distantly supervised open-domain QA (DS-QA) method on publicly available English datasets, even though the target task is not a why-QA.

We propose new BERT-based methods for recognizing event causality such as “smoke cigarettes” –> “die of lung cancer” written in web texts. In our methods, we grasp each annotator’s policy by training multiple classifiers, each of which predicts the labels given by a single annotator, and combine the resulting classifiers’ outputs to predict the final labels determined by majority vote. Furthermore, we investigate the effect of supplying background knowledge to our classifiers. Since BERT models are pre-trained with a large corpus, some sort of background knowledge for event causality may be learned during pre-training. Our experiments with a Japanese dataset suggest that this is actually the case: Performance improved when we pre-trained the BERT models with web texts containing a large number of event causalities instead of Wikipedia articles or randomly sampled web texts. However, this effect was limited. Therefore, we further improved performance by simply adding texts related to an input causality candidate as background knowledge to the input of the BERT models. We believe these findings indicate a promising future research direction.

This paper proposes a neural network-based method for generating compact answers to open-domain why-questions (e.g., "Why was Mr. Trump elected as the president of the US?"). Unlike factoid question answering methods that provide short text spans as answers, existing work for why-question answering have aimed at answering questions by retrieving relatively long text passages, each of which often consists of several sentences, from a text archive. While the actual answer to a why-question may be expressed over several consecutive sentences, these often contain redundant and/or unrelated parts. Such answers would not be suitable for spoken dialog systems and smart speakers such as Amazon Echo, which receive much attention in these days. In this work, we aim at generating non-redundant compact answers to why-questions from answer passages retrieved from a very large web data corpora (4 billion web pages) by an already existing open-domain why-question answering system, using a novel neural network obtained by extending existing summarization methods. We also automatically generate training data using a large number of causal relations automatically extracted from 4 billion web pages by an existing supervised causality recognizer. The data is used to train our neural network, together with manually created training data. Through a series of experiments, we show that both our novel neural network and auto-generated training data improve the quality of the generated answers both in ROUGE score and in a subjective evaluation.

We propose a method for recognizing such event causalities as "smoke cigarettes" → "die of lung cancer" using background knowledge taken from web texts as well as original sentences from which candidates for the causalities were extracted. We retrieve texts related to our event causality candidates from four billion web pages by three distinct methods, including a why-question answering system, and feed them to our multi-column convolutional neural networks. This allows us to identify the useful background knowledge scattered in web texts and effectively exploit the identified knowledge to recognize event causalities. We empirically show that the combination of our neural network architecture and background knowledge significantly improves average precision, while the previous state-of-the-art method gains just a small benefit from such background knowledge.

We propose a semi-supervised learning method for improving why-question answering (why-QA). The key of our method is to generate training data (question-answer pairs) from causal relations in texts such as "[Tsunamis are generated](effect) because [the ocean's water mass is displaced by an earthquake](cause)." A naive method for the generation would be to make a question-answer pair by simply converting the effect part of the causal relations into a why-question, like "Why are tsunamis generated?" from the above example, and using the source text of the causal relations as an answer. However, in our preliminary experiments, this naive method actually failed to improve the why-QA performance. The main reason was that the machine-generated questions were often incomprehensible like "Why does (it) happen?", and that the system suffered from overfitting to the results of our automatic causality recognizer. Hence, we developed a novel method that effectively filters out incomprehensible questions and retrieves from texts answers that are likely to be paraphrases of a given causal relation. Through a series of experiments, we showed that our approach significantly improved the precision of the top answer by 8% over the current state-of-the-art system for Japanese why-QA.

A telemetry signal occurs the line-of-sight problem by the effect of geography in flight test. A fixed relay system or a mobile relay system is used to solve that problem in general but we propose an airborne telemetry relay system using aircraft in this paper. An airborne telemetry relay system receives the telemetry signal of the test vehicle and then retransmits it to the ground system. The receiving antenna which is a phased array antenna can be tilted to beam direction by beam-forming and track the rapidly moving test vehicle in effect. The relay pod which is mounted to an aircraft consists of the front antenna and the side antenna. It receives S-band signal and then down-converts the frequency to L-band signal. As a result, that can remove the frequency interference on an aircraft while retransmitting.

In this paper, we have described a development of the external instrumentation system of a fighter aircraft for flight test. The external instrumentation system consists of the instrumentation pod and the image pod. The instrumentation pod measures a flight data(attitude, velocity, altitude, etc) of the fighter aircraft by using GPS/AHRS sensor. The image pod takes high-speed images for the separation trajectory of a smart bomb with 2 high-speed cameras and video signal for it with one general camera. We have verified the performance of the external instrumentation system through the ground test, the environment test and the flight test.

In this paper, we have described an aircraft captive flight test for the development test of weapon systems. We have conducted a captive flight test for the development of core onboard parts and sensors of airborne weapons and guided missiles. We have used KTX-1/XKO-1 aircraft as a platform for the captive flight test. In order to perform a captive flight test, we have made a captive test pod as a shape of external fuel tank in the XKO-1 and have modified XKO-1 aircraft for a system interface. We have taken a development test about all kinds of seekers, navigation & guidance systems, and core part of guided missile through the aircraft captive flight test.

Event causality knowledge is indispensable for intelligent natural language understanding. The problem is that any method for extracting event causalities from text is insufficient; it is likely that some event causalities that we can recognize in this world are not written in a corpus, no matter its size. We propose a method of hypothesizing unseen event causalities from known event causalities extracted from the web by the semantic relations between nouns. For example, our method can hypothesize "deploy a security camera" -> "avoid crimes" from "deploy a mosquito net" -> "avoid malaria" through semantic relation . Our experiments show that, from 2.4 million event causalities extracted from the web, our method generated more than 300,000 hypotheses, which were not in the input, with 70% precision. We also show that our method outperforms a state-of-the-art hypothesis generation method.

We propose a novel method for acquiring entailment pairs of binary patterns on a large-scale. This method exploits the transitivity of entailment and a self-training scheme to improve the performance of an already strong supervised classifier for entailment, and unlike previous methods that exploit transitivity, it works on a largescale. With it we acquired 138.1 million pattern pairs with 70% precision with such non-trivial lexical substitution as “use Y to distribute X”!“X is available on Y” whose extraction is considered difficult. This represents 50.4 million more pattern pairs (a 57.5% increase) than what our supervised baseline extracted at the same precision.

In this work, we improve the performance of intra-sentential zero anaphora resolution in Japanese using a novel method of recognizing subject sharing relations. In Japanese, a large portion of intrasentential zero anaphora can be regarded as subject sharing relations between predicates, that is, the subject of some predicate is also the unrealized subject of other predicates. We develop an accurate recognizer of subject sharing relations for pairs of predicates in a single sentence, and then construct a subject shared predicate network, which is a set of predicates that are linked by the subject sharing relations recognized by our recognizer. We finally combine our zero anaphora resolution method exploiting the subject shared predicate network and a state-ofthe-art ILP-based zero anaphora resolution method. Our combined method achieved a significant improvement over the the ILPbased method alone on intra-sentential zero anaphora resolution in Japanese. To the best of our knowledge, this is the first work to explicitly use an independent subject sharing recognizer in zero anaphora resolution.

Machine transliteration is a method for automatically converting words in one language into phonetically equivalent ones in another language. Machine transliteration plays an important role in natural language applications such as information retrieval and machine translation, especially for handling proper nouns and technical terms. Four machine transliteration models -- grapheme-based transliteration model, phoneme-based transliteration model, hybrid transliteration model, and correspondence-based transliteration model -- have been proposed by several researchers. To date, however, there has been little research on a framework in which multiple transliteration models can operate simultaneously. Furthermore, there has been no comparison of the four models within the same framework and using the same data. We addressed these problems by 1) modeling the four models within the same framework, 2) comparing them under the same conditions, and 3) developing a way to improve machine transliteration through this comparison. Our comparison showed that the hybrid and correspondence-based models were the most effective and that the four models can be used in a complementary manner to improve machine transliteration performance.

In this paper, we will describe a Korean transliterated foreign word extraction algorithm. In the proposed method, we reformulate the foreign word extraction problem as a syllable-tagging problem such that each syllable is tagged with a foreign syllable tag or a pure Korean syllable tag. Syllable sequences of Korean strings are modelled by Hidden Markov Model whose state represents a character with binary marking to indicate whether the syllable is part of a transliterated foreign word or not. The proposed method extracts a transliterated foreign word with high recall rate and precision rate. Moreover, our method shows good performance even with small-sized training corpora.

We propose a supervised method of extracting event causalities like conduct slash-and-burn agriculture! exacerbate desertification from the web using semantic relation (between nouns), context, and association features. Experiments show that our method outperforms baselines that are based on state-of-the-art methods. We also propose methods of generating future scenarios like conduct slash-and-burn agriculture! exacerbate desertification! increase Asian dust (from China)! asthma gets worse. Experiments show that we can generate 50,000 scenarios with 68% precision. We also generated a scenario deforestation continues! global warming worsens! sea temperatures rise! vibrio parahaemolyticus fouls (water), which is written in no document in our input web corpus crawled in 2007. But the vibrio risk due to global warming was observed in Baker-Austin et al. (2013). Thus, we “predicted” the future event sequence in a sense.

Sequential effects of high glucose on mesangial cell transforming growth factor- β 1 and fibronectin synthesis. Background Transforming growth factor (TGF)- β is recognized as the final common mediator of the principal lesions of diabetic nephropathy such as renal hypertrophy and mesangial expansion. To gain better understanding of the temporal relationships between high glucose (HG) and mesangial cell (MC) TGF- β 1 synthesis and between TGF- β 1 and extracellular matrix (ECM) synthesis, the present study examined early and sequential effects of HG on TGF- β 1 and fibronectin (FN) mRNA expression and protein synthesis. Methods Confluent primary rat MC was stimulated with 5.6 (control) or 30 (high) m m glucose after synchronizing the growth by incubation with serum-free media for 48 hours. Results. Mesangial cell TGF- β 1 mRNA expression increased significantly in six hours and continued to increase until 48 hours in response to HG. The level of TGF- β 1 mRNA was 1.5-fold higher than that of control glucose at six hours and 1.8-fold at 48 hours. TGF- β activity in heat-activated conditioned media under HG increased 1.5- and 1.6-fold at 24 and 48 hours, respectively, compared to control glucose. FN mRNA increased significantly at 24 and 48 hours and 1.4-fold that of control glucose at both time points. FN protein also increased 1.5-fold that of control glucose at 48 hours. Anti-TGF- β antibody completely abolished HG-induced FN synthesis. Conclusions The present findings demonstrate that HG stimulates TGF- β 1 very early and prior to FN production and that HG-induced FN production is mediated by TGF- β . This finding is consistent with the view that TGF- β mediates increased ECM accumulation by MC under high glucose conditions.

We study the storage capacity of a fully connected committee machine with a large number $K$ of hidden nodes. The storage capacity is obtained by analyzing the geometrical structure of the weight space related to the internal representation. By examining the asymptotic behavior of order parameters in the limit of large $K,$ the storage capacity ${\ensuremath{\alpha}}_{c}$ is found to be proportional to $K\sqrt{\mathrm{ln}K}$ up to the leading order. This result satisfies the mathematical bound given by Mitchison and Durbin [Biol. Cybern. 60, 345 (1989)], whereas the replica-symmetric solution in a conventional Gardner approach [Europhys. Lett. 41, 481 (1987); J. Phys. A 21, 257 (1988)] violates this bound.

Inspired by the success of English grapheme-to-phoneme research in speech synthesis, many researchers have proposed phoneme-based English-to-Chinese transliteration models. However, such approaches have severely suffered from the errors in Chinese phoneme-to-grapheme conversion. To address this issue, we propose a new English-to-Chinese transliteration model and make systematic comparisons with the conventional models. Our proposed model relies on the joint use of Chinese phonemes and their corresponding English graphemes and phonemes. Experiments showed that Chinese phonemes in our proposed model can contribute to the performance improvement in English-to-Chinese transliteration.

This paper describes our approach to "NEWS 2009 Machine Transliteration Shared Task." We built multiple transliteration engines based on different combinations of two transliteration models and three machine learning algorithms. Then, the outputs from these transliteration engines were combined using re-ranking functions. Our method was applied to all language pairs in "NEWS 2009 Machine Transliteration Shared Task." The official results of our standard runs were ranked the best for four language pairs and the second best for three language pairs.

This paper proposes a novel framework called bilingual co-training for a large-scale, accurate acquisition method for monolingual semantic knowledge. In this framework, we combine the independent processes of monolingual semantic-knowledge acquisition for two languages using bilingual resources to boost performance. We apply this framework to large-scale hyponymy-relation acquisition from Wikipedia. Experimental results show that our approach improved the F-measure by 3.6--10.3%. We also show that bilingual co-training enables us to build classifiers for two languages in tandem with the same combined amount of data as required for training a single classifier in isolation while achieving superior performance.

Machine transliteration is an automatic method to generate characters or words in one alphabetical system for the corresponding characters in another alphabetical system. There has been increasing concern on machine transliteration as an assistant of machine translation and information retrieval. Three machine transliteration models, including “grapheme-based model”, “phoneme-based model”, and “hybrid model”, have been proposed. However, there are few works trying to make use of correspondence between source grapheme and phoneme, although the correspondence plays an important role in machine transliteration. Furthermore there are few works, which dynamically handle source grapheme and phoneme. In this paper, we propose a new transliteration model based on an ensemble of grapheme and phoneme. Our model makes use of the correspondence and dynamically uses source grapheme and phoneme. Our method shows better performance than the previous works about 15~23% in English-to-Korean transliteration and about 15~43% in English-to-Japanese transliteration.

It is popular in WSD to use contextual information in training sense tagged data. Co-occurring words within a limited window-sized context support one sense among the semantically ambiguous ones of the word. This paper reports on word sense disambiguation of English words using static and dynamic sense vectors. First, context vectors are constructed using contextual words in the training sense tagged data. Then, the words in the context vector are weighted with local density. Using the whole training sense tagged data, each sense of a target word is represented as a static sense vector in word space, which is the centroid of the context vectors. Then contextual noise is removed using a automatic selective sampling. A automatic selective sampling method use information retrieval technique, so as to enhance the discriminative power. In each test case, a automatic selective sampling method retrieves N relevant training samples to reduce noise. Using them, we construct another sense vectors for each sense of the target word. They are called dynamic sense vectors because they are changed according to a target word and its context. Finally, a word sense of a target word is determined using static and dynamic sense vectors. The English SENSEVAL test suit is used for this experimentation and our method produces relatively good results.

There is increasing concern about English-Korean (E-K) transliteration recently. In the previous works, direct converting methods from English alphabets to Korean alphabets were a main research topic. In this paper, we present an E-K transliteration model using pronunciation and contextual rules. Unlike the previous works, our method uses phonetic information such as phoneme and its context. We also use word formation information such as English words of Greek origin, With them, our method shows significant performance increase about 31% in word accuracy.

Developmental capacity of human multipronuclear (PN) zygotes cryopreserved using an ultra-rapid freezing method and electron microscope (EM) grids was studied. Multipronuclear zygotes obtained from a human IVF programme were used as an alternative to normal 2PN zygotes; they were divided into 3PN or >or =4PN zygotes and their in-vitro development and cryo-injury were compared according to PN number. EFS30, which consisted of 30% ethylene glycol, 18% Ficoll, 0. 5 mol/l sucrose and 10% fetal bovine serum with added modified Dulbecco's phosphate buffered saline was used as the freezing solution. After ultra-rapid freezing and thawing 85.5% of multipronuclear zygotes survived. A comparison of cleavage rates between the control and cryopreserved groups showed no significant difference (3PN; 81.3 and 85.4% and > or =4PN; 90.0 and 95.7% respectively). Comparing the in-vitro development after thawing up to blastocyst formation on day 5 after IVF, the outcome of the frozen 3PN group (22.0%) was not different from that of control 3PN group (38.5%), while the outcome of the frozen > or =4PN group (4.5%) was significantly lower than that of control > or =4PN group (44.4%) (P < 0.05).

In recent years, statistical approaches on ATR (Automatic Term Recognition) have achieved good results. However, there are scopes to improve the performance in extracting terms still further. For example, domain dictionaries can improve the performance in ATR. This paper focuses on a method for extracting terms using a dictionary hierarchy. Our method produces relatively good results for this task.