Dialogue Generator Research Articles

Improving Dialogue Management Through Data Optimization

In task-oriented dialogue systems, the ability for users to effortlessly communicate with machines and computers through natural language stands as a critical advancement. Central to these systems is the dialogue manager, a pivotal component tasked with navigating the conversation to effectively meet user goals by selecting the most appropriate response. Traditionally, the development of sophisticated dialogue management has embraced a variety of methodologies, including rule-based systems, reinforcement learning, and supervised learning, all aimed at optimizing response selection in light of user inputs. This research casts a spotlight on the pivotal role of data quality in enhancing the performance of dialogue managers. Through a detailed examination of prevalent errors within acclaimed datasets, such as Multiwoz 2.1 and SGD, we introduce an innovative synthetic dialogue generator designed to control the introduction of errors precisely. Our comprehensive analysis underscores the critical impact of dataset imperfections, especially mislabeling, on the challenges inherent in refining dialogue management processes

Musicking ubuntu/unhu Sustainability Human Rights and Social Justice Activism in the Chimurenga Music of Thomas Mapfumo

During the Second Chimurenga – Zimbabwe Liberation War – music was used to promote social justice and fight the oppressive colonial system. The struggle for justice for all was driven by the spirit of ubuntu/unhu, which embraces the capacity of indigenous African cultures to express humanity, respect, solidarity, and justice relevant to building an inclusive community. Chimurenga music, performed and popularised by Thomas Mapfumo who coined the term in the 1960s, played an important role during the liberation struggle, delivering messages on socio-political concerns, and engaging Zimbabweans on the issues of human rights, corruption, democracy, resistance, and economic transformation. This article discusses the use of Mapfumo’s Chimurenga music in post-colonial Zimbabwe as socio-political discourse and a dialogue generator for sustainability

Answering Student Queries with a Supervised Memory Conversational Agent

This paper describes a discussion-bot that provides answers to students’ questions about the Data Science master program at the University of Lyon 1. Based on a seq2seq architecture combined with a supervised memory module, the bot identifies the questioner’s interest and encodes relevant information from the past conversation to provide personalized answers. A dialogue generator based on hand-crafted dialogues was built to train our model on these synthetic dialogues. The agent and its memory are adaptable to another context by modifying the intention database of the generator. The model was deployed and the results show that the discussion-bot meets most students’ learning requests. We discuss further directions that might be taken to increase the model's effectiveness.

OPAL: Ontology-Aware Pretrained Language Model for End-to-End Task-Oriented Dialogue

AbstractThis paper presents an ontology-aware pretrained language model (OPAL) for end-to-end task-oriented dialogue (TOD). Unlike chit-chat dialogue models, task-oriented dialogue models fulfill at least two task-specific modules: Dialogue state tracker (DST) and response generator (RG). The dialogue state consists of the domain-slot-value triples, which are regarded as the user’s constraints to search the domain-related databases. The large-scale task-oriented dialogue data with the annotated structured dialogue state usually are inaccessible. It prevents the development of the pretrained language model for the task-oriented dialogue. We propose a simple yet effective pretraining method to alleviate this problem, which consists of two pretraining phases. The first phase is to pretrain on large-scale contextual text data, where the structured information of the text is extracted by the information extracting tool. To bridge the gap between the pretraining method and downstream tasks, we design two pretraining tasks: ontology-like triple recovery and next-text generation, which simulates the DST and RG, respectively. The second phase is to fine-tune the pretrained model on the TOD data. The experimental results show that our proposed method achieves an exciting boost and obtains competitive performance even without any TOD data on CamRest676 and MultiWOZ benchmarks.

BEET: Behavior Engagement Emotion Trigger – A AI-based model to maximize audience engagement

Engagement and measuring is a key performance for any content. The wide spread of the Internet and the adoption in our daily life, audience has many options to explore and audience developed the power of rejection. This power of rejection reduces the time of exposure and engagement to any contents including the in person engagement. The access to the online tools and portals provide an escape route from human-tohuman interaction. A new tool to maximize audience engagement to a content that can influence a decision becomes a key performance and an asset for exposure. It enables a better delivery of the Contents, an increased interaction with the content, simplify the process of creating an engaging content, and maximize the exposure time. This paper presents a prototype, and evaluation of the performance of an automated Content Behavior Emotion Engagement Model “BEEM” that uses Deep Learning and Big Data analysis to discover the Behavior Emotion trend of the audience. BEETAIR is a new innovative framework that will transform the media market through Behavior Emotion Engagement Trigger Analyzer, and Intelligent Recommender. It uses The Artificial Intelligent-based Dialogue Generator for Maximum Audience Engagement

Review of State-of-the-Art Design Techniques for Chatbots

Amazon’s Alexa, Apple’s Siri, Google Assistant and Microsoft’s Cortana, clearly illustrate the impressive research work and potentials to be explored in the field of conversational agents. Conversational agent, chatter-bot or chatbot is a program expected to converse with near-human intelligence. Chatbots are designed to be used either as task-oriented ones or simply open-ended dialogue generator. Many approaches have been proposed in this field which ranges from earlier versions of hard-coded response generator to the advanced development techniques in Artificial Intelligence. In a broader sense, these can be categorized as rule-based and neural network based. While rule-based relies on predefined templates and responses, a neural network based relies on deep learning models. Rule-based are preferable for simpler task-oriented conversations. Open-domain conversational modeling is a more challenging area and uses mostly neural network-based approaches. This paper begins with an introduction of chatbots, followed by in-depth discussion on various classical or rule-based and neural-network-based approaches. The evaluation metrics employed for chatbots are mentioned. The paper concludes with a table consisting of recent research done in the field. It covers all the latest and significant publications in the field, the evaluation metrics employed, the corpus which is used as well as the possible areas of enhancement that exist in the proposed techniques.

CERG: Chinese Emotional Response Generator with Retrieval Method.

The dialogue system has always been one of the important topics in the domain of artificial intelligence. So far, most of the mature dialogue systems are task-oriented based, while non-task-oriented dialogue systems still have a lot of room for improvement. We propose a data-driven non-task-oriented dialogue generator “CERG” based on neural networks. This model has the emotion recognition capability and can generate corresponding responses. The data set we adopt comes from the NTCIR-14 STC-3 CECG subtask, which contains more than 1.7 million Chinese Weibo post-response pairs and 6 emotion categories. We try to concatenate the post and the response with the emotion, then mask the response part of the input text character by character to emulate the encoder-decoder framework. We use the improved transformer blocks as the core to build the model and add regularization methods to alleviate the problems of overcorrection and exposure bias. We introduce the retrieval method to the inference process to improve the semantic relevance of generated responses. The results of the manual evaluation show that our proposed model can make different responses to different emotions to improve the human-computer interaction experience. This model can be applied to lots of domains, such as automatic reply robots of social application.

An approach to website schema.org design

Schema.org offers to web developers the opportunity to enrich a website's content with microdata and schema.org. For large websites, implementing microdata can take a lot of time. In general, it is necessary to perform two main activities, for which we lack methods and tools. The first consists in designing what we call the website schema.org, which is the fragment of schema.org that is relevant to the website. The second consists in adding the corresponding microdata tags to the web pages. In this paper, we describe an approach to the design of a website schema.org. The approach consists in using a human–computer task-oriented dialogue, whose purpose is to arrive at that design. We describe a dialogue generator that is domain independent but that can be adapted to specific domains. We propose a set of six evaluation criteria that we use to evaluate our approach and that could be used in future approaches.

An evaluation of multicriteria analysis for DSS in public policy decision

The definition of the routes for the future Adriatic highway along the Adriatic coast has given rise to a great number of discussions, dilemmas, conflicting interests and attitudes, considering the two alternatives of the route, i.e. the coastal and the continental one. Since the ‘classical’ methods (primarily feasibility studies), have yielded almost the same results for both alternatives, the government (the road management authorities) have decided upon the application of multicriteria analysis for finding the optimal alternative of the highway route. Applying multicriteria analysis, i.e. the PROMETHEE method and the program package GAIA, it was attempted to ‘model’ a very complex structure of the problem and by engaging a team of experts to make as objective as possible the evaluation procedure and the weight pondering of each group of criteria which were evidently in conflict. The influence exerted by parameters of each criterion weight on the obtained solution was estimated by different scenarios of weight pondering, and by an analysis of the stability of criterion weights. The paper also presents further development of the methodology used for deciding on alternative solutions of the highway route according to the DSS concept. Consequently, further phases include the integration of the data bases, model bases and the dialogue generator with the graphic interpretation of the results.

Design of Intelligent Dialogue in Decision Support Systems

This article deals with the question of the nature of the interactive dialogue which should take place in any good DSS, with the ways to easily design and implement it (actually leading to the idea of a dialogue generator) and with the features of such dialogues which require the use of Artificial Intelligence to fully play its role of effective interactivity. It stresses the opinion that in the short-term business oriented DSS will benefit less of Artificial Intelligence techniques at the level of its core functionalities than at the level of the interactive dialogue it offers

A structured approach to designing human-computer dialogues

This article presents a conceptual model and software tool for designing and implementing flexible human-computer dialogues. The tool is referred to as a dialogue generator. The rationale for building such a tool is discussed in the context of other work in the literature, and is based on a fully operational version written in APL. A dialogue generator is important because of the cost of designing dialogues, and the rapidly increasing importance of dialogue generation due to the proliferation of interactive applications of computers. A command language model is used for the target dialogue because of its inherent simplicity, which promotes a structured and streamlined approach based on interaction events, which are described in tabular form in the dialogue data-base. “Help” and “abort” facilities are provided, as are facilities for controlling the flow of the dialogue. The implementation is discussed, particularly in terms of how to represent the various types of information pertaining to an interaction event, and how to store and organize the information. The concepts of user guided and system guided dialogues are re-examined in the context of the flow-control mechanism. Both styles of dialogue are needed, depending on the user, and both can be covered by the model described. The model is compared with others in the literature, and its capabilities are evaluated against published rules for dialogue programming.

A dialogue generator

AbstractThis paper describes work done on the Dialogue Generator, DIAGEN, at the Computing Research Centre in Bratiplava. The general aim of this work was to design and implement a generalized software tool for creating an interactive interface which would permit the non‐programmer to work directly with the computer in various application areas.The Dialogue Generator, DIAGEN, separating completely the dialogue from the application program that is to be dialogue‐driven has been implemented. A specialized DIAGEN Language is used for the dialogue description in the form of a scenario. The DIAGEN Conversational System interprets the dialogue program generated from the scenario, and controls the flow of a dialogue session.

A dialogue generator and its use in DSS design

This paper describes a conceptual model and the experimental implementation of a Decision Support System (DSS) dialogue generator. The dialogue generator is a tool for the DSS builder to develop the human/computer interface efficiently and expeditiously. It enables the DSS builder to speed up the initial implementation, provides flexible interfaces, makes it easier to adapt to changing needs, and offers the ability to experiment with prototypes, all of which are factors that contribute to a successful DSS implementation. The paper also provides examples of how the dialogue generator was useful in implementing two experimental systems.