Identification of sections in charge of a request for quotation (RFQ), a type of business-specific document that seeks an itemized list of prices for a product or service, is usually performed manually and is very time-consuming, especially in the manufacturing industry. This study presents a simple semi-supervised classification approach for automatic section identification of RFQ documents. We conceive the identification task as text classification task for different sections and introduce novel features derived from unlabeled data to enhance the performance. We evaluate the usefulness of our approach in a series of experiments on a collection of RFQ documents in the actual business operations and obtain satisfactory results for most test collections.

In this paper, we explore the utility of sentiment analysis and text classification of voice of the customer (VOC) for improving churn prediction, which is a task to detect customers who are about to quit. Our work is motivated by the observation that the increase of customer satisfaction will reproduce churn and the customer satisfaction can be reflected in some degree by applying NLP techniques on VOC, the unstructured textual information which captures a view of customer's attitude and feedbacks. To the best of our knowledge, this is the first work that introduces text classification of VOC to churn prediction task. Experiments show that adding VOC analysis into a conventional churn prediction model results in a significant increase in predictive performance.

Today customers voice attitudes, opinions and their experience about some brands, companies, products or services through center calls, web reviews or SNS and analyzing them is an important task. On the other hand, customer loyalty has long been a topic of high interest in both academia and industry. Therefore, it is attractive to consider exploiting customer loyalty information by analyzing the voice of customer. However, although many previous studies focused on analyzing attitudes, opinions, sentiments of the text data, no work has been conducted from the perspective of customer loyalty, which is reflected by a combination of customer attitudes and behavior. In this work, we present JCLIC, Japanese Customer Loyalty Information Corpus, which is a corpus for analyzing customer loyalty information. For each review we have annotated detailed customer loyalty information which contains: loyalty degree that reflects loyalty level of the customer, loyalty expression that expresses the customer loyalty, loyalty type that indicates the category to which loyalty expression belongs., reason expression that expresses why the customer have such loyalty degree, and reason type that indicates the category to which reason expression belongs.. We describe our annotation scheme and annotation process, present results of an agreement study and give some statistics about the corpus we have annotated.

In order to acquire useful information and knowledge from documents written in natural languages and use them on various applications, the documents first need to be transformed into appropriate forms so that computers can (understand the contents and) handle them more easily. In this paper, transform processings whose usefulness has gained consensus to a certain degree will be called fundamental natural language processings. Typical among them are “morphological analysis” where sentences are segmented into words and each word is assigned a part of speech, and “dependency parsing” where dependency structures between constituent elements such as words and phrases are determined. Evaluative information analysis which we are going to introduce in this paper is also becoming popular as a fundamental natural language processing. It is an analytical processing to judge whether a given expression denotes a positive or negative opinion. We have been developing several systems for fundamental natural language processing. To return what we have gained to the society, we make those systems available to the public through ALAGIN. One of them is our evaluative information analysis system that will be presented in Section 2. Evaluative information analysis In this paper, we describe the fundamental natural language processing tools (evaluative expression analyzer, morphological analyzer, and syntactic parser) that we have developed and released through Advanced Language Information Forum (ALAGIN).

In this paper we explore the utility of sentiment analysis and semantic word classes for improving why-question answering on a large-scale web corpus. Our work is motivated by the observation that a why-question and its answer often follow the pattern that if something undesirable happens, the reason is also often something undesirable, and if something desirable happens, the reason is also often something desirable. To the best of our knowledge, this is the first work that introduces sentiment analysis to non-factoid question answering. We combine this simple idea with semantic word classes for ranking answers to why-questions and show that on a set of 850 why-questions our method gains 15.2% improvement in precision at the top-1 answer over a baseline state-of-the-art QA system that achieved the best performance in a shared task of Japanese non-factoid QA in NTCIR-6.

This paper proposes a method to improve the accuracy of bilingual texts (bitexts) dependency parsing by using an auto-generated bilingual treebank created with the help of statistical machine translation (SMT) systems. Previous bitext parsing methods use human-annotated bilingual treebanks that are costly and troublesome to obtain. In the proposed method, we use an auto-generated bilingual treebank to train the parsing models. First, an SMT system is used to translate a monolingual treebank into the target language; then, a monolingual parser for the target language is used to parse the translated sentences. Since the auto-translated sentences and auto-parsed trees in the auto-generated bilingual treebank are far from perfect, the bilingual constraints are not sufficiently reliable. To overcome this problem, we propose a method to verify the reliability of the constraints using a large amount of target monolingual and bilingual unannotated data. Finally, we design a set of effective bilingual features for parsing models on the basis of the verified constraints. We conduct the experiments using a standard test data. The experimental results show that our bitext parser significantly outperforms monolingual parsers. Moreover, our method is still able to provide improvement when we use a larger monolingual treebank containing over 50 000 sentences. We also test the proposed method with different SMT systems and the results show that our method is very robust to the noise. In particular, the proposed method can be used in a purely monolingual setting with the help of SMT. That is, it does not need the human translation of the test set as previous methods do.

Together with the ever-growing amount of Chinese web data, the number of opinions voiced by Chinese users is rapidly increasing, and analyzing them is an important task. This paper introduces a Chinese Evaluative Information Analyzer (CEIA) and proposes a method to improve its performance. We use evaluative information as a unifying term for the information about attitudes, opinions, sentiments and so on. This paper makes three contributions: (i) CEIA can identify and analyze a more diverse and richer set of evaluative information than previous studies for Chinese; (ii) to implement the system, we constructed an original annotated corpus for Chinese evaluative information and built a large sentiment dictionary; (iii) we introduce syntactic dependency, semantic class and distance features to improve the evaluative information extraction. The performance of the system and the effectiveness of the newly introduced features are evaluated in a series of experiments on our Chinese evaluative information corpus. Title and Abstract in Chinese ¥ ¥' 'µ$f ÂI I¤ ‐ d{X‰˚o˙¥'~iUd†—fi ,{µ$u'˝{jfiU •⁄

We propose a method to improve the accuracy of parsing bilingual texts (bitexts) with the help of statistical machine translation (SMT) systems. Previous bitext parsing methods use human-annotated bilingual treebanks that are hard to obtain. Instead, our approach uses an auto-generated bilingual treebank to produce bilingual constraints. However, because the auto-generated bilingual treebank contains errors, the bilingual constraints are noisy. To overcome this problem, we use large-scale unannotated data to verify the constraints and design a set of effective bilingual features for parsing models based on the verified results. The experimental results show that our new parsers significantly outperform state-of-the-art baselines. Moreover, our approach is still able to provide improvement when we use a larger monolingual treebank that results in a much stronger baseline. Especially notable is that our approach can be used in a purely monolingual setting with the help of SMT.

This paper presents a simple yet effective semi-supervised method to improve Chinese word segmentation and POS tagging. We introduce novel features derived from large auto-analyzed data to enhance a simple pipelined system. The auto-analyzed data are generated from unlabeled data by using a baseline system. We evaluate the usefulness of our approach in a series of experiments on Penn Chinese Treebanks and show that the new features provide substantial performance gains in all experiments. Furthermore, the results of our proposed method are superior to the best reported results in the literature.

In Chinese texts, words composed of single or multiple characters are not separated by spaces, unlike most western languages. Therefore Chinese word segmentation is considered an important first step in machine translation (MT) and its performance impacts MT results. Many factors affect Chinese word segmentations, including the segmentation standards and segmentation strategies. The performance of a corpus-based word segmentation model depends heavily on the quality and the segmentation standard of the training corpora. However, we observed that existing manually annotated Chinese corpora tend to have low segmentation granularity and provide poor morphological information due to the present segmentation standards. In this paper, we introduce a short-unit standard of Chinese word segmentation, which is particularly suitable for machine translation, and propose a semi-automatic method of transforming the existing corpora into the ones that can satisfy our standards. We evaluate the usefulness of our approach on the basis of translation tasks from the technology newswire domain and the scientific paper domain, and demonstrate that it significantly improves the performance of Chinese-Japanese machine translation (over 1.0 BLEU increase).

In this paper, we present a discriminative word-character hybrid model for joint Chinese word segmentation and POS tagging. Our word-character hybrid model offers high performance since it can handle both known and unknown words. We describe our strategies that yield good balance for learning the characteristics of known and unknown words and propose an error-driven policy that delivers such balance by acquiring examples of unknown words from particular errors in a training corpus. We describe an efficient framework for training our model based on the Margin Infused Relaxed Algorithm (MIRA), evaluate our approach on the Penn Chinese Treebank, and show that it achieves superior performance compared to the state-of-the-art approaches reported in the literature.

Performances of a magnetic particle driven micromixer are predicted numerically. This micromixer takes advantages of mixing enhancements induced by alternating actuation of magnetic particles suspended in the fluid. Effects of the magnetic actuation force, switching frequency and channel’s lateral dimension have been investigated. Numerical results show that the magnetic particle actuation at an appropriate frequency causes effective mixing. The optimum switching frequency depends on the channel’s lateral dimension and the applied magnetic force. The optimum switching frequency obtained from the present numerical prediction is in good agreement with the theoretical analysis. The proposed simple mixing scheme not only provides an excellent mixing, even in simple microchannel, but also can be easily applied to “lab-on-a-chip” applications with a pair of external electromagnets.

Performances of a magnetic particle driven micromixer are predicted numerically. This micromixer takes advantages of mixing enhancements induced by alternating actuation of magnetic particles suspended in the fluid. Effects of magnetic actuation force, switching frequency and channel's lateral dimension have been investigated. Numerical results show that the magnetic particle actuation at an appropriate frequency causes effective mixing and the optimum switching frequency depends on the channel's lateral dimension and the applied magnetic force. The maximum efficiency is obtained at a relatively high operating frequency for large magnetic actuation forces and narrow microchannels. If the magnetic particles are actuated with a much higher or lower frequency than the optimum switching frequency, they tend to add limited agitation to the fluid flow and do not enhance the mixing significantly. The optimum switching frequency obtained from the present numerical prediction is in good agreement with the theoretical analysis. The proposed mixing scheme not only provides an excellent mixing, even in simple microchannel, but also can be easily applied to lab-on-a-chip applications with a pair of external electromagnets.

We study the light verb constructions in Japanese, the constructions of expressions with light verb “suru“. Such factors as numerous semantic variants and the syntactic complexity of corresponding Chinese expressions of LVCs put great difficulties in the Japanese-Chinese translation, so current commercially available Japanese-Chinese translation softwares give bad results for LVCs translation. We systematically analyze and describe the phenomena and propose translation rules for the conversion of Japanese LVCs into Chinese. we conducted manual experiments using 200 sentences to evaluate these rules and achieved a translation accuracy of over 80%. We also introduce jaw, a pattern-based translation engine, which can be applied to any target language. We implemented our translation rules for LVCs in the Chinese version, jaw/Chinese and verified their applicability in real MT system by experiments using 100 sentences. Evaluations both by hand and by the machine indicate that it provides high performance and utility.

Performances of a hybrid electrokinetic-passive micromixer are predicted numerically. An h/p-type spectral element method is used to simulate the mixing behavior in microdevices. The numerical algorithm employs modal spectral expansion in quadrilateral and unstructured triangular meshes and provides high-order numerical accuracy. A second-order accurate, stiffly stable integration scheme is used for temporal integration. In the numerical technique, the electric double layer is not resolved to avoid expensive computation, rather a slip velocity is assigned at the channel surface based on the electric field and the electroosmotic mobility. The presented hybrid mixing scheme takes advantages of mixing enhancements induced by asymmetric flow geometries and electrokinetic relay actuation. Effects of relay frequency, applied electric potential, channel width, and channel geometry on micromixing have been conducted. Numerical results show that electrokinetic relay at an appropriate frequency causes effective mixing. Moreover, asymmetric flow geometries and narrow channel width are critical for ultraeffective mixing. The proposed hybrid mixing scheme not only provides excellent mixing within very short time, but also can easily be integrated with microdevices for “lab-on-a-chip” applications because there is no need of any external mechanical pumps.

Numerical study on a rapid micromixer based on hybrid electrokinetic relay and asymmetric serpentine structures is presented. Effective mixing of liquids is essential in many applications such as drug delivery, DNA analysis/sequencing, pheromone synthesis in micro bioreactors, and biological/chemical agent detections. Rapid mixing can reduce the analysis time and permit high throughput in lab-on-a-chip or micro total analysis systems (μTAS). The proposed hybrid mixing takes advantages of both mixing enhancements induced by asymmetric flow geometries and the electrokinetic relay actuating. Simulation results show that the micro mixer is able to achieve high mixing efficiencies (94.3%) in short time (1.2s). Effects of relay frequency, electric field and channel geometry on micro-mixing have been conducted. Numerical results show that electrokinetic relay at an appropriate frequency causes effective micromixing. Moreover asymmetric flow geometries are critical for ultra effective mixing.