教育领域多轮对话机器人的算法设计与实现

随着自然语言处理技术的发展,对话机器人以节省人工和易于嵌入等特点受到业界青睐.为了满足在线教育的需求,本文提出的教育领域多轮对话机器人具备和用户针对知识点进行深入对话的能力.本文介绍了此对话机器人的实现全流程:采取用户模拟器生成教育领域语料,使用意图识别和槽位填充实现自然语言理解,通过对话状态追踪和对话策略设计多轮对话...     

Natural Language Processing for Industry

Recently, natural language processing applications have become very popular in the industry. Examples of such applications include “semantic” enterprise search engines, document categorizers, speech recognizers and – last but not least – conversational agents, also known as virtual assistants or “chatbots”. The latter in particular are very sought-after in the customer care domain, where the aim is to complement the live agent experience with an artificial intelligence able to help users fulfil a task. In this paper, we discuss the challenges and limitations of industrial chatbot applications, with a particular focus on the “human-in-the-loop” aspect, whereby a cooperation between human and machine takes place in mutual interest. Furthermore, we analyse how the same aspect intervenes in other industrial natural language processing applications.     

Designing and evaluating a wizarded uncertainty-adaptive spoken dialogue tutoring system

We describe the design and evaluation of two different dynamic student uncertainty adaptations in wizarded versions of a spoken dialogue tutoring system. The two adaptive systems adapt to each student turn based on its uncertainty, after an unseen human “wizard” performs speech recognition and natural language understanding and annotates the turn for uncertainty. The design of our two uncertainty adaptations is based on a hypothesis in the literature that uncertainty is an “opportunity to learn”; both adaptations use additional substantive content to respond to uncertain turns, but the two adaptations vary in the complexity of these responses. The evaluation of our two uncertainty adaptations represents one of the first controlled experiments to investigate whether substantive dynamic responses to student affect can significantly improve performance in computer tutors. To our knowledge we are the first study to show that dynamically responding to uncertainty can significantly improve learning during computer tutoring. We also highlight our ongoing evaluation of our uncertainty-adaptive systems with respect to other important performance metrics, and we discuss how our corpus can be used by the wider computer speech and language community as a linguistic resource supporting further research on effective affect-adaptive spoken dialogue systems in general.     

面向任务的基于深度学习的多轮对话系统与技术

自然语言是人类智慧的结晶,以自然语言的形式与计算机进行交互是人们长久以来的期待。随着自然语言处理技术的发展与深度学习方法的兴起,人机对话系统成为了新的研究热点。人机对话系统按照功能可以分为任务导向型对话系统、闲聊型对话系统、问答型对话系统。任务导向型对话系统是一种典型的人机对话系统,旨在帮助用户完成某些特定的任务,有着十分重要的学术意义和应用价值。文中系统地阐述了一种在实际工程应用中的任务导向型对话系统的通用框架,主要包括自然语言理解、对话管理以及自然语言生成3个部分;介绍了上述各部分所采用的经典深度学习和机器学习方法。最后,对自然语言理解任务进行了实证性的实验验证与分析,结果表明文中内容可以为任务导向型对话系统的构建提供有效指导。     

Shifting the Design Philosophy of Spoken Natural Language Dialogue: From Invisible to Transparent Systems

Spoken natural language interfaces are characterized by a high degree of opacity. This characteristic leads many new users to converse with the machine on the basis of an inaccurate image of the system, which results in numerous dysfunctions and failures in the human-machine spoken dialogue. The application of the principle of transparency aims to correct this problem by enabling the system to make it “visible” (it can, so to speak, be seen “from the inside”). This report presents a study that compares two versions of the same spoken dialogue system that differ only by some of their dialogue strategies. In particular, one version of this system starts the dialogue with a recall of the main system's function followed by an open-ended prompt, while the other version only produces an open-ended prompt (What can I do for you?). Moreover, one version of this system makes explicit some possible actions in specific dialogue contexts, while the other version leaves these possible actions implicit. These strategies were compared by having a group of 28 users test them in real conditions of use. The results highlight that transparency strategies can help users in formulating simple and acceptable requests, satisfying their goals more easily while provoking less recognition and comprehension errors.     

Gestures that people can understand and use

Recent advances in computing devices push researchers to envision new interaction modalities that go beyond traditional mouse and keyboard input. Typical examples are large displays for which researchers hope to create more “natural” means of interaction by using human gestures and body movements as input. In this article, we reflect about this goal of designing gestures that people can easily understand and use and how designers of gestural interaction can capitalize on the experience of 30 years of research on visual languages to achieve it. Concretely, we argue that gestures can be regarded as “visual expressions to convey meaning” and thus are a visual language. Based on what we have learned from visual language research in the past, we then explain why the design of a generic gesture set or language that spans many applications and devices is likely to fail. We also discuss why we recommend using gestural manipulations that enable users to directly manipulate on-screen objects instead of issuing commands with symbolic gestures whose meaning varies among different users, contexts, and cultures.     

主动联想型对话机器人的设计与实现

人机对话系统通常仅是用户命令的执行者,很少能主动发起对话,或依据对话以外的内容展开个性化的人机交互.本文提出了联想型对话机器人的概念和对应设计方案,即通过融合用户提问和用户个人特征的方式产生对话内容,这样的对话系统不仅可以主动发出对话邀请,也可以个性化地回答用户问题,提升人机对话的满意度.文章首先介绍了联想型对话机器人...     

一种对话机器人开发技术综述与系统架构实现

对话机器人是使用自然语言处理与生成技术,模拟人类对话逻辑并与人进行交流的计算机程序.作为新一代人工智能产品的人机交互主要入口,逻辑实现方式与交互是对话机器人设计的关键.本文结合对话机器人的技术特点,在实现过程中采用基于Python语言的Django微服务Web应用开发框架,将对话算法模型与逻辑处理过程进行微服务化API...     

Data-driven models for timing feedback responses in a Map Task dialogue system

Traditional dialogue systems use a fixed silence threshold to detect the end of users’ turns. Such a simplistic model can result in system behaviour that is both interruptive and unresponsive, which in turn affects user experience. Various studies have observed that human interlocutors take cues from speaker behaviour, such as prosody, syntax, and gestures, to coordinate smooth exchange of speaking turns. However, little effort has been made towards implementing these models in dialogue systems and verifying how well they model the turn-taking behaviour in human–computer interactions. We present a data-driven approach to building models for online detection of suitable feedback response locations in the user's speech. We first collected human–computer interaction data using a spoken dialogue system that can perform the Map Task with users (albeit using a trick). On this data, we trained various models that use automatically extractable prosodic, contextual and lexico-syntactic features for detecting response locations. Next, we implemented a trained model in the same dialogue system and evaluated it in interactions with users. The subjective and objective measures from the user evaluation confirm that a model trained on speaker behavioural cues offers both smoother turn-transitions and more responsive system behaviour.     

Learning perceptually grounded word meanings from unaligned parallel data

In order for robots to effectively understand natural language commands, they must be able to acquire meaning representations that can be mapped to perceptual features in the external world. Previous approaches to learning these grounded meaning representations require detailed annotations at training time. In this paper, we present an approach to grounded language acquisition which is capable of jointly learning a policy for following natural language commands such as “Pick up the tire pallet,” as well as a mapping between specific phrases in the language and aspects of the external world; for example the mapping between the words “the tire pallet” and a specific object in the environment. Our approach assumes a parametric form for the policy that the robot uses to choose actions in response to a natural language command that factors based on the structure of the language. We use a gradient method to optimize model parameters. Our evaluation demonstrates the effectiveness of the model on a corpus of commands given to a robotic forklift by untrained users.     

Robotics software systems

A robotics software “system” is defined here as one which allows robot users to program robot tasks in terms of key states of the task, instead of manipulator motions. It consists of two subsystems: a language system and a planning system. The language system involves the design of syntax and semantics of a robot programming language whereas the planning system determines specific manipulator movements for a given task defined in a task-level language. This paper describes the major components of a robotics software system and reviews principal research findings in the related aspects including programming languages, manipulator and world modelings, motion planning, and graphic simulation. Underlying research issues are addressed at the end.     

Real user evaluation of a POMDP spoken dialogue system using automatic belief compression

This article describes an evaluation of a POMDP-based spoken dialogue system (SDS), using crowdsourced calls with real users. The evaluation compares a “Hidden Information State” POMDP system which uses a hand-crafted compression of the belief space, with the same system instead using an automatically computed belief space compression. Automatically computed compressions are a way of introducing automation into the design process of statistical SDSs and promise a principled way of reducing the size of the very large belief spaces which often make POMDP approaches intractable. This is the first empirical comparison of manual and automatic approaches on a problem of realistic scale (restaurant, pub and coffee shop domain) with real users. The evaluation took 2193 calls from 85 users. After filtering for minimal user participation the two systems were compared on more than 1000 calls.     

Mining language variation using word using and collocation characteristics

Two textual metrics “Frequency Rank” (FR) and “Intimacy” are proposed in this paper to measure the word using and collocation characteristics which are two important aspects of text style. The FR, derived from the local index numbers of terms in a sentences ordered by the global frequency of terms, provides single-term-level information. The Intimacy models relationship between a word and others, i.e. the closeness a term is to other terms in the same sentence. Two textual features “Frequency Rank Ratio (FRR)” and “Overall Intimacy (OI)” for capturing language variation are derived by employing the two proposed textual metrics. Using the derived features, language variation among documents can be visualized in a text space. Three corpora consisting of documents of diverse topics, genres, regions, and dates of writing are designed and collected to evaluate the proposed algorithms. Extensive simulations are conducted to verify the feasibility and performance of our implementation. Both theoretical analyses based on entropy and the simulations demonstrate the feasibility of our method. We also show the proposed algorithm can be used for visualizing the closeness of several western languages. Variation of modern English over time is also recognizable when using our analysis method. Finally, our method is compared to conventional text classification implementations. The comparative results indicate our method outperforms the others.     

Security types preserving compilation

Starting from the seminal work of Volpano and Smith, there has been growing evidence that type systems may be used to enforce confidentiality of programs through non-interference. However, most type systems operate on high-level languages and calculi, and “low-level languages have not received much attention in studies of secure information flow” (Sabelfeld and Myers, [Language-based information-flow security. IEEE Journal on Selected Areas in Communications 2003; 21:5–19]). Therefore, we introduce an information flow type system for a low-level language featuring jumps and calls, and show that the type system enforces termination-insensitive non-interference.Furthermore, information flow type systems for low-level languages should appropriately relate to their counterparts for high-level languages. Therefore, we introduce a compiler from a high-level imperative programming language to our low-level language, and show that the compiler preserves information flow types.     

Modeling in the design of a KBS validation system

We describe and illustrate the modeling issues in the design of a system for validation of knowledge based systems (KBSs). the domain of such a validation system is “KBSs and their validation problems.” the basic idea in our solution is the following. Since different KBSs may use different knowledge representation languages, we first represent the target KBS (i.e., the KBS to be validated) in a general formal model of KBS, and then validate it in this form. the advantage of this strategy is that validation problem solving needs only to refer to the common language of the general formal model. We present a set of possible conceptual abstraction levels in such a model, and argue that each level is associated with a related view on validation problems. Since high level characterizations are difficult to abstract from current knowledge representation languages, we consider the formal aspects of modeling mainly at the “lowest” level, the so-called inference primitive level. We illustrate the approach by formalizing a solution for selected modeling issues at this level.     

Designing habitable dialogues for speech-based interaction with computers

Habitability refers to the match between the language people employ when using a computer system and the language that the system can accept. In this paper, the concept of “habitability” is explored in relation to the design of dialogues for speech-based systems. Two studies investigating the role of habitability in speech systems for banking applications are reported. The first study employed a speech-driven automated teller machine (ATM), using a visual display to indicate available vocabulary. Users made several distinct types of error with this system, indicating that habitability in speech systems cannot be achieved simply by displaying the input language. The second study employed a speech input/speech output home banking application, in which system constraints were indicated by either a spoken menu of words or a “query-style” prompt (e.g. “what service do you require?”). Between-subjects comparisons of these two conditions confirmed that the “menu-style” dialogue was rated as more habitable than the “query-style”. It also led to fewer errors, and was rated as easier to use, suggesting that habitability is a key issue in speech system usability. Comparison with the results of the first study suggests that for speech input, spoken menu prompts may be more habitable than similar menus shown on a visual display. The implications of these results to system design are discussed, and some initial dialogue design recommendations are presented.     

Exception handling in the spreadsheet paradigm

Exception handling is widely regarded as a necessity in programming languages today and almost every programming language currently used for professional software development supports some form of it. However, spreadsheet systems, which may be the most widely used type of “programming language” today in terms of number of users using it to create “programs” (spreadsheets), have traditionally had only extremely limited support for exception handling. Spreadsheet system users range from end users to professional programmers and this wide range suggests that an approach to exception handling for spreadsheet systems needs to be compatible with the equational reasoning model of spreadsheet formulas, yet feature expressive power comparable to that found in other programming languages. We present an approach to exception handling for spreadsheet system users that is aimed at this goal. Some of the features of the approach are new; others are not new, but their effects on the programming language properties of spreadsheet systems have not been discussed before in the literature. We explore these properties, offer our solutions to problems that arise with these properties, and compare the functionality of the approach with that of exception handling approaches in other languages     

Route instructions in map-based human–human and human–computer dialogue: A comparative analysis

When conveying information about spatial situations and goals, speakers adapt flexibly to their addressee in order to reach the communicative goal efficiently and effortlessly. Our aim is to equip a dialogue system with the abilities required for such a natural, adaptive dialogue. In this paper we investigate the strategies people use to convey route information in relation to a map by presenting two parallel studies involving human–human and human–computer interaction. We compare the instructions given to a human interaction partner with those given to a dialogue system which reacts by basic verbal responses and dynamic visualization of the route in the map. The language produced by human route givers is analyzed with respect to a range of communicative as well as cognitively crucial features, particularly perspective choice and references to locations across levels of granularity. Results reveal that speakers produce systematically different instructions with respect to these features, depending on the nature of the interaction partner, human or dialogue system. Our further analysis of clarification and reference resolution strategies produced by human route followers provides insights into dialogue strategies that future systems should be equipped with.     

INDIANA: An interactive system for assisting database exploration

We propose INDIANA, a system conceived to support a novel paradigm of database exploration. INDIANA assists the users who are interested in gaining insights about a database though an interactive and incremental process, like a conversation that does not happen in natural language. During this process, the system iteratively provides the user with some features of the data that might be “interesting” from the statistical viewpoint, receiving some feedbacks that are later used by the system to refine the features provided to the user in the next step. A key ability of INDIANA is to assist “data enthusiastic” users (i.e., inexperienced or casual users) in the exploration of transactional databases in an interactive way. For this purpose, we develop a number of novel, statistically-grounded algorithms to support the interactive exploration of the database. We report an in-depth experimental evaluation to show that the proposed system guarantees a very good trade-off between accuracy and scalability, and a user study that supports the claim that the system is effective in real-world database-exploration tasks.