研究論文

コールセンターなどの対話型サービスで，自然言語処理技術を利用した機械による対話システムの導入が進んでいる．しかしながら，言語タスクの難しさや現在の対話システムの限界から，未だ人手が欠かせない状況である．例えば，旅行情報サイトにおける対話型の宿推薦サービスでは，ユーザが提示する要求に合わせてオペレータが宿を推薦する必要がある．この際，ユーザは日程や人数などの具体的な要求だけでなく，「子連れに優しい宿が良い」などの抽象的な要求1を提示することがある [1]．具体的な要求はルールとして使うことで条件に合致した宿の検索が可能だが，抽象的な要求は曖昧性が高く対処が難しい．抽象的な要求に対してオペレータは，宿の推薦と同時に「こちらの宿は，キッズスペースもあり子連れにオススメです」と推薦根拠も提示することが多い．このような抽象的な要求に対する推薦根拠の提示は，推薦の説得力を増すだけでなく，ユーザのより具体的な要求を引き出すことを可能にする． 本研究では，抽象的な要求に対して根拠を含んだ推薦文を提示する対話システムの実現を目指す．

Universal Dependencies (UD) は、多言語間で共通のアノテーション方式を用いて係り受けのツリーバンクを開発する国際プロジェクトである。浅原ほか (2019) は日本語の UD リソースの現状をまとめている。この中で、元テキストも含めて再配布可能なものは UD Japanese PUD と UD Japanese GSD の 2 つである。しかしながら、これらのリソースも、ライセンスや文の欠損などのさまざまな問題を抱えている。これらの問題を解決するために、我々は UD Japanese GSD の再整備を進めている。ライセンスや失われた情報の復元を進めるとともに、ほかの日本語 UD リソースに合わせて、Omura and Asahara (2018) の手法に基づいたデータの整備を進めた。また、新たに固有表現情報を付与した。これらの作業により、spaCy 標準日本語モデルへの依存構造解析・固有表現抽出モデルの搭載が可能になる。spaCy は多言語の字句解析・固有表現抽出・品詞タグ付け・ラベル付き依存構造解析機能を提供する汎用自然言語処理フレームワークであるが、言語モデル整備時に学習元データを同梱する必要があった。今回、商用利用可能なライセンスに変更し、固有表現情報を付与した UD Japanese GSD を再整備することで spaCyの言語モデル整備に必要な標準的な要件を満たすことになる。

旅行情報サイト「じゃらん net」1には日々レビューが投稿され，さまざまな感想・意見・要望が集積されている．実際に宿泊したカスタマーの目線からなる情報であり，有益な情報源である．これを用いて，どのような宿の特徴がカスタマーに好評であるかという知識を得て整理をすることは，宿を提案する上で有益である．本研究では，1 文を知識の単位として扱い，ある 1 文と含意関係にある複数の文をその 1 文に集約することで知識を整理することを目的とする．そして，その目的 のために文間含意関係認識器の学習用コーパスの構築に取り組む．関係の表現形式は前提文が仮説文を含意するか否かの 2 値2とする．これまでの含意関係コーパス構築ではアノテータに1 文を提示し，事例3が与えた関係ラベルをもつようにもう 1 文を作文4してもらう方式が一般的である．しかしこのような作成方法は高コストであり，また偏った事例を収集してしまう可能性がある．例えば，Tsuchiyaは SNLI コーパス [2] にはもラベルの予測が前提文なしに可能な事例が多数コーパス内に存在することを示した．そこで，本研究ではアノテータによる作文は行わず，自然文のみで含意関係コーパスを構築する．また，応用と分析のしやすいコーパスを目指し，原子的な事例のみを含めるため，複雑な文構造をもつ文は除外し た．加えて，アノテーション作業を確実にし，事例の分析もしやすくするために，非含意の場合はその根拠となる箇所を選択してもらった．著者が知る限りこのような根拠付きのアノテーションは含意関係アノテーションにおいて初めての試みである．以上の特徴をもつ本研究で作成したコーパスは公開を予定しており，アノテーション候補選定のために作成した宿の特徴判定文コーパスや感情極性コーパスも公開する予定である．

Task-oriented dialog systems which assist users to complete tasks like hotel reservation, are drawing great attentions among both research and industry. Compared to conventional pipelined system, recently emerging end-to-end trainable dialog systems are showing many favorable characteristics – because of the neural models that directly learn from chatlogs of human-to-human conversation employed, such systems hold the promise of low data preparation cost, flexible response generation and the ability to evolve with new data. In this work, we are going to explore the possibility to bring this end-to-end trainable framework to the hotel reservation chatbot application, where we encounter two new problems: 1. numerical slots-filling and 2. multi-turn dialog management. To the best of our knowledge, both of them can not be fully solved using currently available end-to-end frameworks. In this paper, we will focus on these two problems and propose possible workarounds which can lead to satisfactory results.

Paraphrases are important linguistic resources for a wide variety of NLP applications. Many techniques for automatic paraphrase mining from general corpora have been proposed. While these techniques are successful at discovering generic paraphrases, they often fail to identify domain-specific paraphrases (e.g., {“staff ”, “concierge”} in the hospitality domain). This is because current techniques are often based on statistical methods, while domain-specific corpora are too small to fit statistical methods. In this paper, we present an unsupervised graph-based technique to mine paraphrases from a small set of sentences that roughly share the same topic or intent. Our system, ESSENTIA, relies on word-alignment techniques to create a word-alignment graph that merges and organizes tokens from input sentences. The resulting graph is then used to generate candidate paraphrases. We demonstrate that our system obtains high quality paraphrases, as evaluated by crowd workers. We further show that the majority of the identified paraphrases are domain-specific and thus complement existing paraphrase databases.

Understanding what makes people happy is a central topic in psychology. Prior work has mostly focused on developing self-reporting assessment tools for individuals and relies on experts to analyze the periodic reported assessments. One of the goals of the analysis is to understand what actions are necessary to encourage modifications in the behaviors of the individuals to improve their overall well-being. In this paper, we outline a complementary approach; on the assumption that the user journals her happy moments as short texts, a system can analyze these texts and propose sustainable suggestions for the user that may lead to an overall improvement in her well-being. We prototype one necessary component of such a system, the Happiness Entailment Recognition (HER) module, which takes as input a short text describing an event, a candidate suggestion, and outputs a determination about whether the suggestion is more likely to be good for this user based on the event described. This component is implemented as a neural network model with two encoders, one for the user input and one for the candidate actionable suggestion, with additional layers to capture psychologically significant features in the happy moment and suggestion. Our model achieves an AU-ROC of 0.831 and outperforms our baseline as well as the current state-of-the-art Textual Entailment model from AllenNLP by more than 48% of improvements, confirming the uniqueness and complexity of the HER task.

Online users are constantly seeking experiences, such as a hotel with clean rooms and a lively bar, or a restaurant for a romantic rendezvous. However, e-commerce search engines only support queries involving objective attributes such as location, price, and cuisine, and any experiential data is relegated to text reviews. In order to support experiential queries, a database system needs to model subjective data. Users should be able to pose queries that specify subjective experiences using their own words, in addition to conditions on the usual objective attributes. This paper introduces OpineDB, a subjective database system that addresses these challenges. We introduce a data model for subjective databases. We describe how OpineDB translates subjective queries against the subjective database schema, which is done by matching the user query phrases to the underlying schema. We also show how the experiential conditions specified by the user can be combined and the results aggregated and ranked. We demonstrate that subjective databases satisfy user needs more effectively and accurately than alternative techniques through experiments with real data of hotel and restaurant reviews.

We describe our experience in developing ConciergeBot, an industrial strength question-answering bot for hotels. The bot automatically suggests answers for information-seeking questions over an input knowledge base of facts about the hotel and its amenities. We demonstrate how ConciergeBot handles unique challenges that arise in our setting. More specifically, we show how our system trains effective models with limited training data, how it can be deployed in different hotels with almost no hotel-specific tuning, and how it manages heterogeneity in questions and data. Our experiments validate that ConciergeBot achieves high precision (78%) and good recall (71%) with as few as 1,300 questions for training purposes.

Open Information Extraction (OPENIE) extracts meaningful structured tuples from freeform text. Most previous work on OPENIE considers extracting data from one sentence at a time. We describe NEURON, a system for extracting tuples from question-answer pairs. Since real questions and answers often contain precisely the information that users care about, such information is particularly desirable to extend a knowledge base with. NEURON addresses several challenges. First, an answer text is often hard to understand without knowing the question, and second, relevant information can span multiple sentences. To address these, NEURON formulates extraction as a multi-source sequence-to-sequence learning task, wherein it combines distributed representations of a question and an answer to generate knowledge facts. We describe experiments on two real-world datasets that demonstrate that NEURON can find a significant number of new and interesting facts to extend a knowledge base compared to state-of-the-art OPENIE methods.