Dialogue Corpus Research Articles

Dialogic interaction between player and non-player characters in animal crossing

Abstract Dialogic interaction is a distinctive feature of Animal Crossing, a social simulation video game developed by Nintendo, yet, little attention has been paid to it from a discourse analytical perspective. This paper aims to explore how AC characters are characterised through language and which discourse strategies are applied to engage players. The analysis, based on a corpus of dialogues transcribed by fans of the game, relies on corpus-linguistics methodologies and can be framed within the context of ludolinguistics. The study shows that emotive language is used to create an active interaction between player and non-player characters (NPCs). Even though NPCs are minimally characterized in terms of gender, age, or social status, the collocational analysis of “I” and “you” highlights two opposite personalities interacting in the dialogues: type A, lexically represented as extroverted, dynamic, and active, and type B, represented as kind, hesitant, and passive.

Reinforcing personalized persuasion in task-oriented virtual sales assistant.

Existing task-oriented virtual agents can assist users with simple tasks like ticket booking, hotel reservations, etc. effectively and with high confidence. These virtual assistants, however, assume specific, predictable end-user behavior, such as predefined/servable objectives, which results in conversation failures in challenging situations, such as when goals are unavailable. Inspired by the practice and its efficacy, we propose an end-to-end framework for task-oriented persuasive dialogue generation that combines pre-training and reinforcement learning for generating context-aware persuasive responses. We utilize four novel rewards to improve consistency and repetitiveness in generated responses. Additionally, a meta-learning strategy has also been utilized to make the model parameters better for domain adaptation. Furthermore, we also curate a personalized persuasive dialogue (PPD) corpus, which contains utterance-level intent, slot, sentiment, and persuasion strategy annotation. The obtained results and detailed analysis firmly establish the effectiveness of the proposed persuasive virtual assistant over traditional task-oriented virtual assistants. The proposed framework considerably increases the quality of dialogue generation in terms of consistency and repetitiveness. Additionally, our experiment with a few shot and zero-shot settings proves that our meta-learned model learns to quickly adopt new domains with a few or even zero no. of training epochs. It outperforms the non-meta-learning-based approaches keeping the base model constant. To the best of our knowledge, this is the first effort to improve a task-oriented virtual agent's persuasiveness and domain adaptation.

Построение и визуализация обобщенного диалогового графа по корпусу диалогов

A definition of a generalized dialogue graph is proposed to describe the structure of a dialogue in terms of a corpus of homogeneous dialogues. The task of constructing such a graph is relevant in modern conversational artificial intelligence, however, there are few works with meaningful results, often a complete description of the algorithms is not given and the code with the implementation is not published. In the paper, a method for constructing a generalized dialogue graph is proposed, which is implemented in the Python programming language and made publicly available. Experiments were carried out on open data and the results were described.

Dialogue-adaptive language model pre-training from quality estimation☆

Pre-trained language models (PrLMs) have achieved great success on a wide range of natural language processing tasks by virtue of the universal language representation ability obtained by self-supervised learning on a large corpus. These models are pre-trained on standard plain texts with general language model (LM) training objectives, which would be insufficient to model dialogue-exclusive attributes like specificity and informativeness reflected in these tasks that are not explicitly captured by the pre-trained universal language representations. In this work, we propose dialogue-adaptive pre-training objectives (DAPO) derived from quality estimation to simulate dialogue-specific features, namely coherence, specificity, and informativeness. As the foundation for model pre-training, we synthesize a new dialogue corpus and build our training set with two unsupervised methods: 1) coherence-oriented context corruption, including utterance ordering, insertion, and replacement, to help the model capture the coherence inside the dialogue contexts; and 2) specificity-oriented automatic rescoring, which encourages the model to measure the quality of the synthesized data for dialogue-adaptive pre-training by considering specificity and informativeness. Experimental results on widely used open-domain response selection and quality estimation benchmarks show that DAPO significantly improves the baseline models and achieves state-of-the-art performance on the MuTual leaderboard, verifying the effectiveness of estimating quality evaluation factors into pre-training.

DESIGNING EFFECTIVE CHATBOT SYSTEM USING GRU WITH BEAM SEARCH

Artificial Intelligence (AI) based Chatbot is a moderately new technology in the world. AI and Natural Language Processing (NLP) empowers a Chatbot to converse like a human being. Chatbots have become popular recently as they diminish human efforts by automating various tasks. AI-based Chatbot learns from the previous discussion and generates an appropriate response or action for the input given by the user. In the proposed research work we designed AI-based Chatbot system using the Sequence to Sequence (Seq2Seq) model. This system uses a Gated Recurrent Unit (GRU) for encoder and decoder. In the proposed model the GRU encoder accepts a query from the user. The GRU encoder uses an attention mechanism to consider only relevant information and convert it into the context vector form. A context vector is another input to the GRU decoder. The GRU decoder generates a response using the Beam search algorithm. The research work uses Cornell Movie Dialogue Corpus to train the proposed interactive Chatbot system. It is observed that the proposed model with the combination of GRU and Beam search gives better accuracy with the minimum loss for testing data. These experimental results are better than existing approaches that use LSTM Seq2Seq models to train Chatbot systems.

Automatic offline annotation of turn-taking transitions in task-oriented dialogue

As the volume of recorded conversations continues to surge, so does the need for their automatic processing. Plenty of information beyond words may be extracted from the speech signal that could be valuable in domains such as call-center quality assurance. In particular, describing the dynamics of turn-taking exchanges allows for a deeper understanding of the development and outcome of a dialogue. In this paper, we investigate the construction of an automatic turn-taking annotation tool based on recordings of entire conversations (in offline mode) — an unexplored topic to our knowledge. We experiment with two supervised learning approaches, using recurrent neural networks and random forests, on a corpus of Argentine Spanish task-oriented dialogues annotated with 12 turn-taking categories following standard guidelines. Our models achieve promising results, with F1 scores ranging 0.7–0.9 for the most frequent labels (e.g., smooth switches, backchannels), but much lower for the least frequent ones (various kinds of interruptions), for which further research is needed. We also evaluate our best-performing models considering their generalizability in scenarios of growing difficulty, including dialogues in two different languages (English and Slovak). Finally, to address the typical data scarcity issue, we analyze the impact of combining training data from different corpora, again including cross-linguistic data.

A rich task-oriented dialogue corpus in Vietnamese

A rich task-oriented dialogue corpus in Vietnamese

Can a robot laugh with you?: Shared laughter generation for empathetic spoken dialogue.

Spoken dialogue systems must be able to express empathy to achieve natural interaction with human users. However, laughter generation requires a high level of dialogue understanding. Thus, implementing laughter in existing systems, such as in conversational robots, has been challenging. As a first step toward solving this problem, rather than generating laughter from user dialogue, we focus on “shared laughter,” where a user laughs using either solo or speech laughs (initial laugh), and the system laughs in turn (response laugh). The proposed system consists of three models: 1) initial laugh detection, 2) shared laughter prediction, and 3) laugh type selection. We trained each model using a human-robot speed dating dialogue corpus. For the first model, a recurrent neural network was applied, and the detection performance achieved an F1 score of 82.6%. The second model used the acoustic and prosodic features of the initial laugh and achieved a prediction accuracy above that of the random prediction. The third model selects the type of system’s response laugh as social or mirthful laugh based on the same features of the initial laugh. We then implemented the full shared laughter generation system in an attentive listening dialogue system and conducted a dialogue listening experiment. The proposed system improved the impression of the dialogue system such as empathy perception compared to a naive baseline without laughter and a reactive system that always responded with only social laughs. We propose that our system can be used for situated robot interaction and also emphasize the need for integrating proper empathetic laughs into conversational robots and agents.

Towards personalized persuasive dialogue generation for adversarial task oriented dialogue setting

In recent years, task-oriented virtual assistants have gained huge popularity and demand in both research and industry communities. The primary aim of a task-oriented dialogue agent is to assist end-users in accomplishing a task successfully and satisfactorily. Existing virtual agents have acquired proficiency in assisting users in solving simple tasks such as restaurant bookings. However, they operate under the deterministic presumption that end-users will have a servable task objective, which makes them inadequate under adversarial situations such as goal unavailability. On the other hand, human agents accomplish users’ tasks even in many goal unavailability scenarios by persuading them towards a similar goal to the user’s proposed task. Motivated by the limitation, the current work proposes and builds a novel transformer-based context-aware personalized persuasive virtual assistant (CoPersUasive VA), which also serves end-users in task unavailability situations. The proposed CoPersUasive VA recognizes goal conflicts through user sentiment and identifies an appropriate persuasion strategy using ongoing dialogue context and user personality. Depending on users’ proposed goals, it finds a similar servable goal and persuades them with the identified persuasion strategy. The obtained experimental results and detailed post-analysis firmly establish that the proposed model effectively enhances the capability of task-oriented virtual assistants to deal with the task failures caused by goal unavailability. The obtained findings also suggest the crucial role of dialogue context in identifying an appropriate and appealing persuasion strategy. The proposed CoPersUasive model could easily be adapted to any other domain by fine-tuning the model on an underlying task. Furthermore, we developed a personalized persuasive multi-intent conversational dialogue corpus annotated with intent, slot, sentiment, and dialogue act for electronic domain.22The dataset and code will be made available to accelerate research towards the development of persuasive virtual sales assistants.

Hierarchical latent variables structure for topic aware multi-turn conversation

Hierarchical structures have emerged as a powerful framework for response generation, which can generate fluent responses in multi-turn conversation. However, the generated responses are often generic and bland. Some researchers have adopted latent variables to improve the diversity of responses, but they can not make full use of the information from multi-turn background, leading to meaningless replies with irrelevant topics. In order to fully utilize the background information for generating diverse and informative responses, we propose a Variational Hierarchical Conversation RNNs model with Topic aware latent variables (VHCR-T). The model contains three levels of latent variables: the global level latent variable to represent background information, the topic level latent variable to capture topic-related information, and the sentence level latent variable to increase the response diversity. When modeling the topic information, we design two different topic level latent variables to maintain the dialog coherence and role preference, and to enhance the context sensitiveness, respectively. Experimental results on Cornell Movie Dialog and Ubuntu Dialog Corpus show that our model outperforms the state-of-the-art models for multi-turn conversation generation in terms of diversity and informativeness, verifying the effectiveness of our VHCR-T model.

Understanding Laughter in Dialog

This work explores laughter within a corpus of three-party, task-based dialogs with native and non-native speakers of English, each consisting of two players and a facilitator, in relation to whether the laughter is perceived as serving discourse functions or rather as genuinely mirthful according to a small number of annotators’ (2) inspection of a substantical multimodal dialog corpus (18 interactions of approximately 10 min each). We test the hypothesis that those different types of laughter have occurrence patterns that relate in different ways to the topical structure of the conversations, with discourse laughter showing a stronger tendency to occur at topic termination points. All laughter events (569) are assigned to one of three values, discourse, mirthful or ambiguous, and are studied with respect to their distribution across the dialog topic sections. The analysis explores interactions among laughter type and section type, also with respect to other variables such as the facilitators’ feedback and the speakers’ conversational role and gender. Discourse laughter is more frequent at topic termination points than at topic beginnings, also in comparison to mirthful laughter. Discourse laughter is also highly associated with facilitators’ feedback type, especially at topic ends. Finally, there are few distinctive effects of gender, and an interaction among speaker role and laughter type. The results strengthen the hypothesis of the discourse function of laughter, indicating a systematicity in discourse laughter, in that it is more predictable and highly associated with the dialog topic termination points, and, on the contrary, a less systematic distribution of mirthful laughter, which shows no particular pattern in relation to topic boundaries.

‘けど’로 종결되는 ‘미완결발화표현’ - politeness이론과의 연관성에 대하여 -

This study analyzes the relationship between the Suspended Clauses and the function of politeness theory in the free dialogue of the Meidai Dialogue Corpus that is a natural discourse. As a result, the function of providing information that fulfills the function of preface, topic presentation, and opinion expression by the speaker's utterance function of 'KEDO' has a function as the speaker's positive politeness strategy. In addition, it becomes clear that the function of requesting information, such as supplementing, correcting explanations, expressing caution, and requesting expressions, is related to the speaker's negative politeness strategy. In this way, the politeness discourse function by ‘KEDO’ seems to be at a higher level when compared with the basic utterance function ‘KEDO’ is uttered in discourse. Further analysis of the relationship with the politeness theory is the main issue for future discourse study, and it is necessary to deepen the consideration and enhance the study.

Simulating and Modeling the Risk of Conversational Search

In conversational search, agents can interact with users by asking clarifying questions to increase their chance of finding better results. Many recent works and shared tasks in both natural language processing and information retrieval communities have focused on identifying the need to ask clarifying questions and methodologies of generating them. These works assume that asking a clarifying question is a safe alternative to retrieving results. As existing conversational search models are far from perfect, it is possible and common that they could retrieve/generate bad clarifying questions. Asking too many clarifying questions can also drain a user’s patience when the user prefers searching efficiency over correctness. Hence, these models can backfire and harm a user’s search experience due to these risks from asking clarifying questions. In this work, we propose a simulation framework to simulate the risk of asking questions in conversational search and further revise a risk-aware conversational search model to control the risk. We show the model’s robustness and effectiveness through extensive experiments on three conversational datasets — MSDialog, Ubuntu Dialog Corpus, and Opendialkg — in which we compare it with multiple baselines. We show that the risk-control module can work with two different re-ranker models and outperform all of the baselines in most of our experiments.

Modeling Japanese Praising Behavior by Analyzing Audio and Visual Behaviors

Praising behavior is considered to be verbal and nonverbal behaviors that expresses praise the behavior and character of the target. However, how one should use verbal and nonverbal behaviors to successfully praise a target has not been clarified. Therefore, we focus on attempts to analyze praising behavior in Japanese dialogue using verbal and nonverbal behaviors. In this study, we attempted to analyze the relationship between praising skills and human behaviors in Japanese dialogue by focusing on voice, head, and face behaviors. First, we created a new dialogue corpus in Japanese containing voice, head, and face behaviors from individuals giving praise (praiser) and receiving praise (receiver), as well as the degree of success of praising (praising score). Second, we developed machine learning models that uses features related to voice, head, and face behaviors to estimate praising skills to clarify which features of the praiser and receiver are important for estimating praising skills. Evaluation resulte showed that some audio features of the praiser are particularly important for estimation of praising skills. Our analysis results demonstrated the importance of features related to the zero-crossing rate, MFCCs of the praiser. Analyzing the features of high importance revealed that the praiser should praise with specific words that mean amazing or great in Japanese and the voice quality of the praiser is considered to be important for praising successfully.

Construction and Analysis of a Dialogue Corpus with Paraphrase Pairs of Indirect and Direct Responses

Construction and Analysis of a Dialogue Corpus with Paraphrase Pairs of Indirect and Direct Responses

Design and release of a multimodal dialogue corpus

Design and release of a multimodal dialogue corpus

A persona aware persuasive dialogue policy for dynamic and co-operative goal setting

Contextualization:In recent years, the popularity of virtual agents particularly task-oriented dialogue agents has increased immensely due to their effectiveness and simplicity in various domains such as industry, e-commerce, and health. Problem:In real-world, users do not have always a predefined and immutable goal, i.e., they may upgrade/downgrade/update their task goal dynamically depending upon their utility and the serving capability of the assisting agent. However, the existing Virtual Agents (VAs) in the dialogue literature relinquish and yield dialogue failure if they find any dynamic goal setting or goal unavailability scenarios. Contributions and methodology:Motivated by these inabilities of existing VAs, we propose some intelligent and expert Dialogue Agents (A Unified Dialogue Agent and Multi-agent Dialogue system) that can deal with dynamic and goal unavailability situations to elevate both user satisfaction and the agent’s utility particularly task success rate. The proposed architecture incorporates a goal guiding module namely Dynamic and Co-Operative Goal Driven Module (DyCoGDM), which traces goal status and resolves goal discrepancy through dynamic goal setting (Goal Formulator) and personalized persuasion (Goal Persuader) mechanisms. We also created and annotated a dialogue corpus because of unavailability of such corpus featured with dynamic and goal unavailability scenarios. Findings and implications:Our proposed method outperforms several baselines and state of the art methods in all evaluation metrics. The proposed VA is capable of dealing with dynamic goals and goal unavailability scenarios effectively. The study found that the persona aware persuasive dialogue agent outperforms generalized persuasive dialogue agent by a large margin. Furthermore, we also observed that the task oriented reward is the most essential reward for training a reinforcement learning based agent and agents trained without task based reward do not even converge.

Enhance the Language Ability of Humanoid Robot NAO through Deep Learning to Interact with Autistic Children

Autism spectrum disorder (ASD) is a life-long neurological disability, and a cure has not yet been found. ASD begins early in childhood and lasts throughout a person’s life. Through early intervention, many actions can be taken to improve the quality of life of children. Robots are one of the best choices for accompanying children with autism. However, for most robots, the dialogue system uses traditional techniques to produce responses. Robots cannot produce meaningful answers when the conversations have not been recorded in a database. The main contribution of our work is the incorporation of a conversation model into an actual robot system for supporting children with autism. We present the use a neural network model as the generative conversational agent, which aimed at generating meaningful and coherent dialogue responses given the dialogue history. The proposed model shares an embedding layer between the encoding and decoding processes through adoption. The model is different from the canonical Seq2Seq model in which the encoder output is used only to set-up the initial state of the decoder to avoid favoring short and unconditional responses with high prior probability. In order to improve the sensitivity to context, we changed the input method of the model to better adapt to the utterances of children with autism. We adopted transfer learning to make the proposed model learn the characteristics of dialogue with autistic children and to solve the problem of the insufficient corpus of dialogue. Experiments showed that the proposed method was superior to the canonical Seq2sSeq model and the GAN-based dialogue model in both automatic evaluation indicators and human evaluation, including pushing the BLEU precision to 0.23, the greedy matching score to 0.69, the embedding average score to 0.82, the vector extrema score to 0.55, the skip-thought score to 0.65, the KL divergence score to 5.73, and the EMD score to 12.21.

Dialogue History Matters! Personalized Response Selection in Multi-Turn Retrieval-Based Chatbots

Existing multi-turn context-response matching methods mainly concentrate on obtaining multi-level and multi-dimension representations and better interactions between context utterances and response. However, in real-place conversation scenarios, whether a response candidate is suitable not only counts on the given dialogue context but also other backgrounds, e.g., wording habits, user-specific dialogue history content. To fill the gap between these up-to-date methods and the real-world applications, we incorporate user-specific dialogue history into the response selection and propose a personalized hybrid matching network (PHMN). Our contributions are two-fold: (1) our model extracts personalized wording behaviors from user-specific dialogue history as extra matching information; (2) we perform hybrid representation learning on context-response utterances and explicitly incorporate a customized attention mechanism to extract vital information from context-response interactions so as to improve the accuracy of matching. We evaluate our model on two large datasets with user identification, i.e., personalized Ubuntu dialogue Corpus (P-Ubuntu) and personalized Weibo dataset (P-Weibo). Experimental results confirm that our method significantly outperforms several strong models by combining personalized attention, wording behaviors, and hybrid representation learning.

Response type selection for chat-like spoken dialog systems based on LSTM and multi-task learning

We propose a method of automatically selecting appropriate responses in conversational spoken dialog systems by explicitly determining the correct response type that is needed first, based on a comparison of the user’s input utterance with many other utterances. Response utterances are then generated based on this response type designation (back channel, changing the topic, expanding the topic, etc.). This allows the generation of more appropriate responses than conventional end-to-end approaches, which only use the user’s input to directly generate response utterances. As a response type selector, we propose an LSTM-based encoder–decoder framework utilizing acoustic and linguistic features extracted from input utterances. In order to extract these features more accurately, we utilize not only input utterances but also response utterances in the training corpus. To do so, multi-task learning using multiple decoders is also investigated.To evaluate our proposed method, we conducted experiments using a corpus of dialogs between elderly people and an interviewer. Our proposed method outperformed conventional methods using either a point-wise classifier based on Support Vector Machines, or a single-task learning LSTM. The best performance was achieved when our two response type selectors (one trained using acoustic features, and the other trained using linguistic features) were combined, and multi-task learning was also performed.