Dialogue and Conversational Agents

A conversational system needs to know how to switch between topics to continue the conversation for a more extended period. For this topic detection from dialogue corpus has become an important task for a conversation and accurate prediction of conversation topics is important for creating coherent and engaging dialogue systems. In this paper, we proposed a topic detection approach with Parallel Latent Dirichlet Allocation (PLDA) Model by clustering a vocabulary of known similar words based on TF-IDF scores and Bag of Words (BOW) technique. In the experiment, we use K-mean clustering with Elbow Method for interpretation and validation of consistency within-cluster analysis to select the optimal number of clusters. We evaluate our approach by comparing it with traditional LDA and clustering technique. The experimental results show that combining PLDA with Elbow method selects the optimal number of clusters and refine the topics for the conversation.

Because ethos is an unavoidable component of dialogue and forms the basis for believing and being persuaded by another's speech, it is an important topic for AI researchers. This paper examines the concept of ethos, especially Aristotle's notions of situated and invented ethos, as it functions in oral and written discourse and then explores what happens to ethos in computer-mediated human-to-human and human-to-machine discourse. The paper draws a number of conclusions that may be of value to researchers in these fields. In particular, it argues that the rhetorical concept of ethos furnishes a broader theoretical framework for understanding design and ethical issues involved in agent credibility than does the artistic notion of believability. The paper concludes by suggesting some nonartistic methods for making agents more credible within the framework of situated ethos.

Understanding the landscape of potential harms from algorithmic systems enables practitioners to better anticipate consequences of the systems they build. It also supports the prospect of incorporating controls to help minimize harms that emerge from the interplay of technologies and social and cultural dynamics. A growing body of scholarship has identified a wide range of harms across different algorithmic technologies. However, computing research and practitioners lack a high level and synthesized overview of harms from algorithmic systems arising at the micro-, meso-, and macro-levels of society. We present an applied taxonomy of sociotechnical harms to support more systematic surfacing of potential harms in algorithmic systems. Based on a scoping review of computing research (n=172), we identified five major themes related to sociotechnical harms-representational, allocative, quality-of-service, interpersonal harms, and social system/societal harms-and sub-themes. We describe these categories and conclude with a discussion of challenges and opportunities for future research.

With the growing popularity of conversational agents based on large language models (LLMs), we need to ensure their behaviour is ethical and appropriate. Work in this area largely centres around the 'HHH’ criteria - making outputs more helpful and honest, and avoiding harmful (biased, toxic, or inaccurate) statements. Whilst this semantic focus isu seful when viewing LLM agents as mere mediums or output-generating systems, it fails to account for pragmatic factors that can make the same speech act seem more or less tactless or inconsiderate in different social situations. With the push towards agentic AI, wherein systems become increasingly proactive in chasing goals and performing actions in the world, considering the pragmatics of interaction becomes essential. We propose an interactional approach to ethics that is centred on relational and situational factors. We explore what it means for a system, as a social actor, to treat an individual respectfully in a (series of) interaction(s). Our work anticipates a set of largely unexplored risks at the level of situated social interaction, and offers practical suggestions to help agentic LLM technologies treat people well.

The dialogical character of Wittgenstein's Philosophical Investigations has received scant attention in the literature, given the work's status in his total oeuvre, and is dismissed as a marginal as compared to the other differences between the Tractatus and the Investigations. The main lines of interpretation that have been proposed see dialogue as a rhetorical technique intended to present erroneous positions and then refute them, as an exemplification of what can be expressed in language (McGinn 1997; Rhees 1998), or as a reflection of Wittgenstein's informal teaching method (Malcolm 2001; Savickey et al. 1990). The present article adopts the perspective that Wittgenstein's use of dialogue makes it possible to track the various modes of language-acts, consonant with his directions to examine the daily use of language (Wittgenstein 2009, §116 and esp. §132), "when language is, as it were, idling." In his later inquiries, Wittgenstein frequently considers the nature of mental states, accompanied by an attempt to characterize the differences between them while at the same time dealing with the cases in which it is difficult to distinguish them. In this process he made a variety of uses of dialogue, each of which embodies a different aspect of language action. Subsequently I will demonstrate that these different uses are not haphazard. A scrutiny of the nature of the dialogue can help us understand the nature of the activity carried out of the state of consciousness. Finally, I propose a distinction among three main types of dialogue: technical, conversational, and reflexive.

Other-repetitions are a device involving the reproduction by a speaker of what another speaker has just said. This paper proposes a solution to automatically detect other-repetitions in French conversational dialogue. A first step of the proposed system consists in finding all possible other-repetitions in the dialogue. A second step is used to select other-repetitions which need to be kept by combining rules with speaker statistics. This automatic detection, evaluated on a one hour dialogue, shows good results according to the expected objectives : recall is 1, and precision is about 80%.

In this paper we describe a design approach for an Artificial Conversational Companion according to earlier identified requirements of utility, adaptivity, conversational capabilities and long-term interaction. The Companion is aimed to help advanced learners of a foreign language to practice conversation via instant messenger dialogues. In order to model a meaningful long-term interaction with an Artificial Conversational Companion for this application case, it is necessary to understand how natural long-term interaction via chat between human language experts and language learners works. For this purpose, we created a corpus from instant messenger-based interactions between native speakers of German and advanced learners of German as a foreign language. We used methods from conversation analysis to identify rules of interaction. Examples from our data set are used to illustrate how particular requirements for the agent can be fulfilled. Finally, we outline how the identified patterns of interaction can be used for the design of an Artificial Conversational Companion.

In this position paper we consider temporal phenomena in interaction with text-based conversational agents. In particular, we focus on two dimensions of time in instant messaging dialogues: responsiveness as a measure for interaction placed in time, and interaction management performed by interaction participants caused by partner's exceeding of the maximum expected responsiveness.

In this position paper we consider temporal phenomena in interaction with text-based conversational agents. In particular, we focus on two dimensions of time in instant messaging dialogues: responsiveness as a measure for interaction placed in time, and interaction management performed by interaction participants caused by partner’s exceeding of the maximum expected responsiveness. Author Keywords

Abstract: In this paper we describe a design approach for an Artificial Conversational Companion according to ear-lier identified requirements of utility, adaptivity, conversational capabilities and long-term interaction. The Companion is aimed to help advanced learners of a foreign language to practice conversation via instant mes-senger dialogues. In order to model a meaningful long-term interaction with an Artificial Conversational Companion for this application case, it is necessary to understand how natural long-term interaction via chat between human language experts and language learners works. For this purpose, we created a corpus from instant messenger-based interactions between native speakers of German and advanced learners of German as a foreign language. We used methods from conversation analysis to identify rules of interaction. Examples from our data set are used to illustrate how particular requirements for the agent can be fulfilled. Finally, we outline how the iden...

In this paper we describe a design approach for an Artificial Conversational Companion according to earlier identified requirements of utility, adaptivity, conversational capabilities and long-term interaction. The Companion is aimed to help advanced learners of a foreign language to practice conversation via instant messenger dialogues. In order to model a meaningful long-term interaction with an Artificial Conversational Companion for this application case, it is necessary to understand how natural long-term interaction via chat between human language experts and language learners works. For this purpose, we created a corpus from instant messenger-based interactions between native speakers of German and advanced learners of German as a foreign language. We used methods from conversation analysis to identify rules of interaction. Examples from our data set are used to illustrate how particular requirements for the agent can be fulfilled. Finally, we outline how the identified patterns of interaction can be used for the design of an Artificial Conversational Companion.

This paper is based on several attempts to provide a definition of an Artificial Companion that can be found in the referenced literature. Although accepted by the research community, such descriptions set very high expectations of such agents, but they do not address the technical feasibility and the system limitations, that is why they need an elaborated and precise explanation. In this paper, we focus on computer agents that simulate human language behaviour, and are aimed to serve, to assist and to accompany their owner over a long period of time, that we call Artificial Conversational Companions. This is the first approach to analyse the effort required to meet the expectations of Artificial Conversational Companion and discuss the identified design issues.

In this position paper we consider temporal phenomena in interaction with text-based conversational agents. In particular, we focus on two dimensions of time in instant messaging dialogues: responsiveness as a measure for interaction placed in time, and interaction management performed by interaction participants caused by partner's exceeding of the maximum expected responsiveness.

This paper is based on several attempts to provide a definition of an Artificial Companion that can be found in the referenced literature. Although accepted by the research community, such descriptions set very high expectations of such agents, but they do not address the technical feasibility and the system limitations, that is why they need an elaborated and precise explanation.

In this paper we describe a design approach for an Artificial Conversational Companion according to earlier identified requirements of utility, adaptivity, conversational capabilities and long-term interaction. The Companion is aimed to help advanced learners of a foreign language to practice conversation via instant messenger dialogues. In order to model a meaningful long-term interaction with an Artificial Conversational Companion for this application case, it is necessary to understand how natural long-term interaction via chat between human language experts and language learners works. For this purpose, we created a corpus from instant messenger-based interactions between native speakers of German and advanced learners of German as a foreign language. We used methods from conversation analysis to identify rules of interaction. Examples from our data set are used to illustrate how particular requirements for the agent can be fulfilled. Finally, we outline how the identified patte...

L’era digitale ha indebolito la capacità di intrattenere una conversazione con sé stessi e con gli altri. Le più effimere forme di storytelling hanno preso il posto del dialogo a quattr’occhi e i piccoli momenti di noia sono avvertiti come un fastidio piuttosto che come un’occasione per l’introspezione. La tecnologia non può insegnare all’uomo a essere umano, ma può creare le condizioni per generare esperienze estremamente dinamiche, creative e arricchenti, che gli consentano di riacquisire le abilità di riflessione e di relazione. Per farlo, come in un processo di scaffolding, deve essere messa da parte al momento giusto. Questa proposta etica di disimpegno trova nella scuola, uno dei contesti di socializzazione nei quali si educa alla vita, un ambiente ideale per la sua realizzazione. Nella speranza di inaugurare una nuova fase della progettazione didattica, il contributo presenta un agente conversazionale il cui obiettivo è quello di abilitare forme di dialogo tra il docente e gli studenti.

Mass media sources, specifically the news media, have traditionally informed us of daily events. In modern times, social media services such as Twitter provide an enormous amount of user-generated data, which have great potential to contain informative news-related content. For these resources to be useful, we must find a way to filter noise and only capture the content that, based on its similarity to the news media, is considered valuable. However, even after noise is removed, information overload may still exist in the remaining data-hence; it is convenient to prioritize it for consumption. To achieve prioritization, information must be ranked in order of estimated importance considering three factors. First, the temporal prevalence of a particular topic in the news media is a factor of importance, and can be considered the media focus (MF) of a topic. Second, the temporal prevalence of the topic in social media indicates its user attention (UA). Last, the interaction between the social media users who mention this topic indicates the strength of the community discussing it, and can be regarded as the user interaction (UI) toward the topic. We propose an unsupervised framework-SociRank-which identifies news topics prevalent in both social media and the news media, and then ranks them by relevance using their degrees of MF, UA, and UI. Our experiments show that SociRank improves the quality and variety of automatically identified news topics.

A conversational system needs to know how to switch between topics to continue the conversation for a more extended period. For this topic detection from dialogue corpus has become an important task for a conversation and accurate prediction of conversation topics is important for creating coherent and engaging dialogue systems. In this paper, we proposed a topic detection approach with Parallel Latent Dirichlet Allocation (PLDA) Model by clustering a vocabulary of known similar words based on TF-IDF scores and Bag of Words (BOW) technique. In the experiment, we use K-mean clustering with Elbow Method for interpretation and validation of consistency within-cluster analysis to select the optimal number of clusters. We evaluate our approach by comparing it with traditional LDA and clustering technique. The experimental results show that combining PLDA with Elbow method selects the optimal number of clusters and refine the topics for the conversation.

Recent innovative practices in man-machine interaction can be achieved through intelligent conversations in the form of stories and storytelling between the human users and the conversational agent. Doing do calls for two major steps: text understanding and text generation. The current study focuses on the first as it aims to provide a description of interactional moves and strategies between teacher and students within the context of human-to-human storytelling. Storytelling sessions in reading classes at different grade levels were audio recorded. transcribed, and analyzed using Graesser, Person and Huber (GPH) scheme (1992). Results reveal that the pattern of exchanges between teachers and students during storytelling sessions is almost exclusively Question and Answer while the strategies employed are those that reveal activating, prompting, hinting, feedbacking, and evaluating. This will help inform human-machine interaction specifically involving intelligent conversational agents in a storytelling context. This is envisioned to later help in the development of AI-based platforms and paradigms within and outside the classroom not only for storytelling purposes, but also for reading and other language-related pedagogical needs.

Keynote at the University of Amsterdam, The Netherlands, 22 November 2018, Digital Communications Methods Lab’s special session on ‘Trust in AI, Social Robots and Conversational Agents’:
http://www.digicomlab.eu/events/current-events/special-session-trust-in-ai-social-robots-and-conversational-agents-with-dr-huma-shah/

The demand for conversational Artificial Intelligence (CAI) is increasing to address customer service solutions [1]. Recent advances in speech recognition technology has accelerated the embedding of personal digital assistants across a range of use-cases in the digital economy, including artificial humans in e-retail, and virtual nurses for healthcare.  Early Conversational Artificial Intelligence (CAI) was experienced through Joseph Weizenbaum’s natural language understanding system Eliza [2]. Eliza’s question-answer format, based on a psychotherapist, allowed humans to interact with the dialogue agent [3]. A similar format emerged in a text-based simulation of schizophrenia through Colby et al.’s 1971 PARRY system [4] helping to train psychiatrists [5].

E-commerce introduced the question-answer paradigm through text-based avatars augmenting online keyword search. One well known is Anna (Fig1) for Swedish furniture company, Ikea’s website. A 24/7 accessible virtual agent, Anna produced a return on investment of 200%; increased sales from Ikea’s digital catalogue by 10%, and reduced Ikea’s call centre costs by 20% [6]. Since 2011 an increase in the use of digital assistants followed the launch of Apple’s Siri personal assistant on its iPhone6 smart ‘phone. 

Does the fun of using conversational AI empower our trust in the convenience that personal digital assistants, such as Amazon’s Alexa, afford? Could conversational artificial intelligence be the next scandal in the digital economy, or could these technologies help to keep our identities safe, our personal data private and our trust maintained? These are important questions as we develop more business and social AI agents, and robots with human language capability to interact with us in shopping malls, airports, hospitals, in our homes as information assistants, carers or companions, and as more businesses transform their processes ready for 5G and keep competitive in the fourth industrial revolution.

References
[1] Peart, A. 2018. Conversational AI is Becoming More Mainstream As Demand Increases. Datanami, August 21, 2018. Accessed from: https://www.datanami.com/2018/08/21/conversational-ai-is-becoming-more-mainstream-as-demand-increases/?platform=hootsuite
[2] Weizenbaum. J. (1966). ELIZA – A Computer Programme for the Study of Natural Language Communication between Men and Machines. Communications of the ACM, 9, pp 36-45
[3] Shah, H., Warwick, K., Vallverdú, J. and Wu, D. 2016. Can Machines Talk? Comparison of Eliza with Modern Dialogue Systems. Computers in Human Behavior, 58, pp. 278-295 10.1016/j.chb.2016.01.004
[4] Colby, K.M., Weber, S.,& Hilf, F.D. (1971). Artificial Paranoia. Artificial Intelligence Vol. 2, pp 1-25
[5] Heiser, J.F., Colby, K. M., Fraught, W.S. and Parkison, R.C. (1979). Can Psychiatrists Distinguish a Computer Simulation of Paranoia from the Real Thing?: The Limitation of Turing-like Tests as Measures of the Adequacy of Simulations. Journal of Psychiatric Research. Vol. 15(3):  pp 149-16
[6] Shah, H. 2005. Text-based dialogical e-query systems: gimmick or convenience? Invited talk in Inaugural colloquium on conversational systems. Digital World Research Centre, University of Surrey: November 25

In this paper we describe a design approach for an Artificial Conversational Companion according to earlier identified requirements of utility, adaptivity, conversational capabilities and long-term interaction. The Companion is aimed to help advanced learners of a foreign language to practice conversation via instant messenger dialogues. In order to model a meaningful long-term interaction with an Artificial Conversational Companion for this application case, it is necessary to understand how natural long-term interaction via chat between human language experts and language learners works. For this purpose, we created a corpus from instant messenger-based interactions between native speakers of German and advanced learners of German as a foreign language. We used methods from conversation analysis to identify rules of interaction. Examples from our data set are used to illustrate how particular requirements for the agent can be fulfilled. Finally, we outline how the identified patterns of interaction can be used for the design of an Artificial Conversational Companion.

In this position paper we consider temporal phenomena in interaction with text-based conversational agents. In particular, we focus on two dimensions of time in instant messaging dialogues: responsiveness as a measure for interaction placed in time, and interaction management performed by interaction participants caused by partner's exceeding of the maximum expected responsiveness.

Text classification has become a critical step in big data analytics. For supervised machine learning approaches to text classification, availability of sufficient training data with classification labels attached to individual text units is essential to the performance. Since labeled data are usually scarce, however, it is always desirable to devise a semi-supervised method where unlabeled data are used in addition to labeled ones. A solution is to apply a latent factor model to generate clustered text features and use them for text classification. The main thrust of the current research is to extend Latent Dirichlet Allocation (LDA) for this purpose by considering word weights in sampling and maintaining balances of topic distributions. A series of experiments were conducted to evaluate the proposed method for classification tasks. The result shows that the topic distributions generated by the balance weighted topic modeling method add some discriminative power to feature generations for classification.