Conversational Robot Research Articles

Neural personalized response generation as domain adaptation

One of the most crucial problem on training personalized response generation models for conversational robots is the lack of large scale personal conversation data. To address the problem, we propose a two-phase approach, namely initialization then adaptation, to first pre-train an optimized RNN encoder-decoder model (LTS model) in a large scale conversational data for general response generation and then fine-tune the model in a small scale personal conversation data to generate personalized responses. For evaluation, we propose a novel human aided method, which can be seen as a quasi-Turing test, to evaluate the performance of the personalized response generation models. Experimental results show that the proposed personalized response generation model outperforms the state-of-the-art approaches to language model personalization and persona-based neural conversation generation on the automatic evaluation, offline human judgment and the quasi-Turing test.

Multiple-Robot Conversational Patterns for Concealing Incoherent Responses

Conversational robots, which are used in the fields of education, therapy, and in expositions, are expected to keep a user engaged in conversation. However, these robots sometimes utter comments that are irrelevant topic to the current context owing to a failure in recognizing the human user’s speech or intention. Such a sudden topic shift is considered to interfere with what we call the sense of conversation with which a person can feel as if he or she is participating in a conversation. In this paper, to reduce the interferes of the sudden topic shift, we propose to use multiple robots in a conversation, in which even an actually irrelevant, sudden topic shift sounds involving possible relevance to be shared with subjects in the ongoing conversation. To verify it, we conducted an experiment in which subjects experienced a conversation with either one or two robots and then evaluated their impression of the conversations. The experimental results showed that the subjects who talked with two robots felt less ignored by the robots, and had less difficulty in continuing the conversation with them, than those who talked with a single robot. Further analysis considering subjects’ social skills raised the possibility of an additional effect on robot coherence perception. Finally, we discuss a new disruption-tolerant conversational system design using multiple robots based on the experimental results.

Intelligent Chatter Bot for Regulation Search

AbstractThis communication presents a functional prototype, named PTAH, implementing a linguistic model focused on regulations in Spanish. Its global architecture, the reasoning model and short statistics are provided for the prototype. It is mainly a conversational robot linked to an Expert System by a module with many intelligent linguistic filters, implementing the reasoning model of an expert. It is focused on bylaws, regulations, jurisprudence and customized background representing entity mission, vision and profile. This Structure and model are generic enough to self-adapt to any regulatory environment, but as a first step, it was limited to an academic field. This way it is possible to limit the slang and data numbers. The foundations of the linguistic model are also outlined and the way the architecture implements the key features of the behavior.

Four-participant group conversation: A facilitation robot controlling engagement density as the fourth participant

In this paper, we present a framework for facilitation robots that regulate imbalanced engagement density in a four-participant conversation as the forth participant with proper procedures for obtaining initiatives. Four is the special number in multiparty conversations. In three-participant conversations, the minimum unit for multiparty conversations, social imbalance, in which a participant is left behind in the current conversation, sometimes occurs. In such scenarios, a conversational robot has the potential to objectively observe and control situations as the fourth participant. Consequently, we present model procedures for obtaining conversational initiatives in incremental steps to harmonize such four-participant conversations. During the procedures, a facilitator must be aware of both the presence of dominant participants leading the current conversation and the status of any participant that is left behind. We model and optimize these situations and procedures as a partially observable Markov decision process (POMDP), which is suitable for real-world sequential decision processes. The results of experiments conducted to evaluate the proposed procedures show evidence of their acceptability and feeling of groupness.

Feature Extraction Method for Clock Drawing Test

Recently, the number of elderly persons with dementia has been increasing. In the past, we proposed a dementia evaluation system using daily conversations and developed the system with a conversational robot. However, the current system is not ready for practical use because it can only evaluate time/geographical orientation and short-term memory, and some methods to evaluate other orientations and functions is required as well. In this paper, we discuss a new dementia evaluation system using not only daily conversations but also drawing tests. The authors employed a Clock Drawing Test (CDT) as a new dementia evaluation test and implemented it in a tablet device. This paper discusses a feature extraction and recognition method to distinguish normal cases from dementia cases. After evaluation experiments, the proposed method could recognize 87.6% of the clock drawing images.

Nonstrict Hierarchical Reinforcement Learning for Interactive Systems and Robots

Conversational systems and robots that use reinforcement learning for policy optimization in large domains often face the problem of limited scalability. This problem has been addressed either by using function approximation techniques that estimate the approximate true value function of a policy or by using a hierarchical decomposition of a learning task into subtasks. We present a novel approach for dialogue policy optimization that combines the benefits of both hierarchical control and function approximation and that allows flexible transitions between dialogue subtasks to give human users more control over the dialogue. To this end, each reinforcement learning agent in the hierarchy is extended with a subtask transition function and a dynamic state space to allow flexible switching between subdialogues. In addition, the subtask policies are represented with linear function approximation in order to generalize the decision making to situations unseen in training. Our proposed approach is evaluated in an interactive conversational robot that learns to play quiz games. Experimental results, using simulation and real users, provide evidence that our proposed approach can lead to more flexible (natural) interactions than strict hierarchical control and that it is preferred by human users.

Older People’s Prior Robot Attitudes Influence Evaluations of a Conversational Robot

As the population ages, healthcare robots may help meet increasing demands for mental and physical health services. However more understanding is required of how to make robots acceptable to older people. This study aimed to assess how older peoples’ robot attitudes and drawings were related to their reactions to a conversational robot. We also assessed whether altering the robot’s virtual face affected peoples’ responses. Twenty participants aged over 55 conversed with a Peoplebot robot for 30 min. During the interaction the robot displayed six different face conditions on its monitor in a randomized order. The six robot conditions varied on two dimensions; (i) facial appearance (humanlike, machinelike, or no face), and (ii) robot gender. Measures included the robot attitudes scale, drawings of a robot’s face prior to the interaction, blood pressure (BP), heart rate, and evaluations of the robot. Results suggest participants did not evaluate the robot’s six face displays conditions differently. However, there was a trend for men to evaluate the robot more highly than women did. Participants’ positive attitudes towards robots before the robot interactions were associated with positive robot evaluations after the interactions. Larger drawings were associated with higher systolic BP after interacting with the robot. These findings suggest that, at least in the short-term, people’s pre-existing mental models of robots may be more important for acceptance than the human or machinelikeness, or even the presence of a robot’s virtual face. More research is needed on gender differences in reactions to eldercare robots. Compared with creating different robot faces to meet individual preferences, promoting positive attitudes towards robots may be a cost-effective method of promoting robot acceptance. Drawings of robots may be a useful, more implicit way of assessing anxiety towards robots in potential users.

Conversational robots: An approach to conversation protocol issues that utilizes the paralinguistic information available in a robot-human setting

Conversational robots: An approach to conversation protocol issues that utilizes the paralinguistic information available in a robot-human setting

Temporal Awareness in Teleoperation of Conversational Robots

Awareness of time is particularly important for teleoperation of conversational robots, both for controlling the robot and for estimating interaction success, because people have a low tolerance for long pauses in conversation. Findings have shown that people engaged in high-workload tasks tend to underestimate the passage of time. This study confirms that this problem exists for operators controlling a conversational robot, and it investigates mechanisms for improving temporal awareness and task performance while minimizing workload. In a laboratory experiment, two approaches to helping an operator perform various information input tasks were compared: first, assisting temporal awareness by using a clock display, and second, using autonomy to assist one of the operator's time-dependent tasks. Results revealed that assisting the task itself, even without the clock, improved not only task performance but also the operator's temporal awareness. However, results regarding the effect of the clock were ambiguous: it increased workload in general and did not help temporal awareness overall, but it did improve temporal awareness for text entry tasks in particular. As text entry is an important task for teleoperation of social robots, we further investigated the problem of improving temporal awareness during text entry tasks. As the first experiment suggested the effectiveness of a clock, we further validated that the clock is specifically useful to improve temporal awareness. These results showed that the clock did not increase workload for text entry tasks; however, for touch-typing operators, the results suggested that showing a clock after the end of an interaction, rather than continuously throughout the task, could lower the operator's perceived workload.

Teleoperation of Multiple Social Robots

Teleoperation of multiple robots by a single operator has been studied extensively for applications such as search and navigation; however, this concept has never been applied to the field of social, conversational robots. In this paper, we explore the unique challenges posed by the remote operation of multiple social robots, where an operator must perform auditory multitasking to assist multiple interactions at once. It describes the general system requirements in four areas: social human-robot interaction (HRI) design, autonomy design, multirobot coordination, and teleoperation interface design. Based on this design framework, we have developed a system in which a single operator can simultaneously control four robots in conversational interactions with users. Key elements of our implementation include a control architecture enabling the scripting of conditional behavior flows for social interaction, a graphical interface enabling an operator to control one robot at a time while monitoring several others in the background, and a technique called “proactive timing control,” which is an automated method for smoothly interleaving the demands of multiple robots for the operator's attention. We also present metrics for describing and predicting robot performance, and we show experimental results demonstrating the effectiveness of our system through simulations and a laboratory experiment based on real-world interactions.

USING TAGS IN AN AIML-BASED CHATTERBOT TO IMPROVE ITS KNOWLEDGE

Nowadays, it is common to find on the Internet different conversational robots which interact with users simulating a natural language conversation. Among them, we can emphasize the chatterbots based on AIML language. In this paper we present an AIML based chatterbot that shows as its main contribution the use of tags and folksonomies. Thanks to its use, we can generate a context for each conversation, being able to maintain a state for each user in the system, and improving the adaptation capabilities of the bot.

Adding TQ-BOT into a Third-party Learning Management System

Intelligent Tutoring Systems are computer programs that aim at providing personalized instruction to students. In recent years, artificial intelligence conversational robots, usually known as chatterbots, have become very popular in the Internet. In this paper we show how chatterbots can be integrated in e-Learning Systems. To perform such an integration the Service Oriented Architecture paradigm is adopted and e-learning standardization initiatives are considered. A middleware is provided to enable the integration and reuse of chatterbots by e-Learning systems enabling a tight control of their operation. Such middleware takes to account several issues such as authorising users, creating instances, transferring data to and from the chatterbot, assigning permissions to users, and subscribing to events. Our approach is applied to the specific case of TQ-Bot, which is use to track and supervise the student progress and to provide answers orienting the student to the more appropriate course contents.

A multi-expert model for dialogue and behavior control of conversational robots and agents

This paper presents an intelligence model for conversational service robots. It employs modules called experts, each of which is specialized to execute certain kinds of tasks such as performing physical behaviors and engaging in dialogues. Some of the experts take charge in understanding human utterances and deciding robot utterances or actions. The model enables switching and canceling tasks based on recognized human intentions, as well as parallel execution of several tasks. This model specifies the interface that an expert must have, and any kind of expert can be employed if it conforms to the interface. This feature makes the model extensible.

TQ-Bot: An AIML-based Tutor and Evaluator Bot

Intelligent Tutoring Systems are computer programs that aim at providing personalized instruction to students. In recent years, conversational robots, usually known as chatterbots, become very popular in the Internet, and ALICE (Artificial Linguistic Internet Computer Entity) is probably the most popular one. ALICE brain is written in AIML (Artificial Intelligence Markup Language), an open XML language. We have considered the combination of both approaches, i.e, the use of AIML-based bots for tutoring purposes in open e-Learning platforms such as Claroline or Moodle. With that aim in mind, we have developed a bot (chatterbot) for helping the students during their learning process and for supporting the activities of the teacher. This bot (TQ-Bot) is able to analyse the requests made by the learners in written natural language and to provide adequate and domain specific answers orienting the student to the right course contents. Besides, TQ-Bot is able to track and supervise the student progress by means of personalized questionnaires. This bot has been developed and integrated as user-friendly modules in Claroline and Moodle.

1P1-F08 英会話ロボットと人間との感性コミュニケーション

In this paper we clarify the factor of the scenario that the English conversation robot should have and factor of the impression after it talks. The factor analysis and text Mining are used for the analysis. The data structure to be analyzed is clarified by classifying the targeted text data into the group of the one which the content is near, and showing the word used in the feature in the attribute such as another of the classified group and character ages statistically.

Domain knowledge query conversation bots in instant messaging (IM)

In this paper, we examine the use of knowledge query technology as applied to conversation bots in the instant messaging environment. Hence, we designed an artificial intelligent conversation robot or bots called Artificial Intelligence Natural language Identity (hereafter, AINI) to mimic human conversation. Our goal is to introduce a Domain Matrix Knowledge Model and an Automated Knowledge Extraction Agent (AKEA) to create AINI’s knowledge bases, and in turn provide intelligent query mechanisms. We report an evaluation on the collection and analysis of a corpus containing over 3280 utterances in a series of real instant messages exchanged between the AINI conversation bot and 65 online “buddies”. About 1721 utterances were produced by AINI, 88.03% were from open-domain knowledge, 2.15% from domain-specific knowledge base and 9.82% were inappropriate and amusing responses. These results show that domain knowledge plays significant roles in conversations between two or more human users and in human–machine conversation.

2P1-H07 未来型住居を想定したロボット群によるホームサービスの試み(空間知)

It reports on the appearance of the demonstration concerning the home service with multiple robots of spreading table, conversation robot, and media mirror in near future house though it was open to the public at the end of last year in Waseda University WABOT-HOUSE laboratory.

Anthropo-morphic conversational robot : Multimodal human interface with para-linguistic information expressing/understanding abilities.

Anthropo-morphic conversational robot : Multimodal human interface with para-linguistic information expressing/understanding abilities.

An automated approach to online digital reference: the Open University Library OPAL Project

Online Personal Academic Librarian (OPAL) is an Open University research project looking at the development of a fully automated 24/7 online system designed to respond to routine and repeat enquiries from Open University distance learners. Looks at the desk and field research conducted to date, and is divided into five sections. First, the reason and need for such an automated question‐answer service are considered. Then examines the current digital reference climate, and a range of possible solutions and technologies is described, including the use of real‐time technologies such as chat and Web contact software, and also the use of fully automated online conversational robots. Then looks at enquirer behaviour and enquiry patterns, drawing on data collected through field research and a data analysis of user e‐mail enquiries. Further research has also revealed the importance of the user’s context and the need for user profiling in delivering an appropriate response to the user, and this is briefly discussed. Concludes with a description of the OPAL prototype to date, and looks ahead to some possible future directions.