• Contenu principal
  • Menu
OpenEdition Books
  • Accueil
  • Catalogue de 15898 livres
  • Éditeurs
  • Auteurs
  • Facebook
  • X
  • Partager
    • Facebook

    • X

    • Accueil
    • Catalogue de 15898 livres
    • Éditeurs
    • Auteurs
  • Ressources numériques en sciences humaines et sociales

    • OpenEdition
  • Nos plateformes

    • OpenEdition Books
    • OpenEdition Journals
    • Hypothèses
    • Calenda
  • Bibliothèques

    • OpenEdition Freemium
  • Suivez-nous

  • Lettre d’information
OpenEdition Search

Redirection vers OpenEdition Search.

À quel endroit ?
  • Accademia University Press
  • ›
  • Collana dell'Associazione Italiana di Li...
  • ›
  • Proceedings of the Seventh Italian Confe...
  • ›
  • Contributed Papers
  • ›
  • Natural Language Generation in Dialogue ...
  • Accademia University Press
  • Accademia University Press
    Accademia University Press
    Informations sur la couverture
    Table des matières
    Liens vers le livre
    Informations sur la couverture
    Table des matières
    Formats de lecture

    Plan

    Plan détaillé Texte intégral 1. Introduction 2. A Dialogue Corpus for Customer-care Domain 3. Balancing information sources in NLG for DS 4. Conclusion and Future Work Bibliographie Notes de bas de page Auteurs

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Ce livre est recensé par

    Précédent Suivant
    Table des matières

    Natural Language Generation in Dialogue Systems for Customer Care

    Mirko Di Lascio, Manuela Sanguinetti, Luca Anselma, Dario Mana, Alessandro Mazzei, Viviana Patti et Rossana Simeoni

    p. 151-156

    Résumé

    In this paper we discuss the role of natural language generation (NLG) in modern dialogue systems (DSs). In particular, we will study the role that a linguistically sound NLG architecture can have in a DS. Using real examples from a new corpus of dialogue in customer-care domain, we will study how the non-linguistic contextual data can be exploited by using NLG.

    Remerciements

    The work of Mirko Di Lascio, Alessandro Mazzei, Manuela Sanguinetti e Viviana Patti has been partially funded by TIM s.p.a. (Studi e Ricerche su Sistemi Conversazionali Intelligenti, CENF_CT_RIC_19_01).

    Texte intégral Bibliographie Notes de bas de page Auteurs

    Texte intégral

    1. Introduction

    1In this paper we present the first results of an ongoing project on the design of a dialogue system for customer care in the telco field. In most of the dialogue systems (DSs), the generation side of the communication is quite limited to the use of templates (Van Deemter, Krahmer, and Theune 2005). Templates are pre-compiled sentences with empty slots that can be filled with appropriate fillers. Most of commercial DSs, following the classical cascade architecture Image 10000000000001D9000000114BE99CA67A42FE0C.jpg(McTear, Callejas, and Griol 2016), use machine learning-based Natural Language Understanding (NLU) techniques to identify important concepts (e.g., intent and entities in (Google 2020)) that will be used by the dialogue manager (i) to update the state of the system and (ii) to produce the next dialogue act (Bobrow et al. 1977; Traum and Larsson 2003), possibly filling the slots in the generation templates.

    2This classical, and quite common, information flow/architecture for dialogue processing has, as a working hypothesis, the assumption that most of necessary information is provided by the user’s utterance: we call this information linguistic channel (L-channel). However, especially in the customer-care domain, this assumption is only partially true. For instance, in the sentence “Scusami ma vorrei sapere come mai mi vengono fatti certi addebiti?” (“Excuse me, I’d like to know why I’m charged certain fees?”), even a very advanced NLU module can produce only a vague information about the user’s request to the DialogueManager. Indeed, in order to provide good enough responses, the DialogueManager resorts to other two sources of information: the domain context channel (DC-channel) and the user model channel (UM-channel). The DC-channel is fundamental to produce the content of the answer, while the UM-channel is necessary to give also the correct form.

    3It is worth noting that both channels, that are often neglected in the design of commercial DSs for customer-care domain, have central roles in the design of (linguistically sound) natural language generation (NLG) systems (Reiter and Dale 2000). In particular, considering the standard architecture for data-to-text NLG systems (Reiter 2007; Gatt and Krahmer 2018), the analysis of the DC-channel exactly corresponds to the content selection task and the UM-channel influences both the sentence planning and sentence realization phases. In other words, the central claims of this paper are that in commercial DSs for customer care: (1) L-channel is often not informative enough and one needs to use the DC-channel and the UM-channel for producing a sufficiently good answer, (2) DC-channel and UM-channel can be exploited by using standard symbolic1 NLG techniques and methods. The remainder of the paper supports both of these claims while presenting our ongoing project on the development of a rule-based NLG prototype to be used in a customer care domain. Section 2 presents the corpus developed in the first stage of this project, consisting of real dialogues containing explanation requests in telco customer-care domain. Section 3 presents an NLG architecture for managing the L-DC-UM channels that can be adopted in a DS for customer care. Finally, Section 4 concludes the paper with few remarks on the current state of the project and on future work.

    2. A Dialogue Corpus for Customer-care Domain

    4This study builds upon the analysis of a corpus of dialogues between customers and a DS for customer service developed by an Italian telecommunications company. The dialogues, which take place by means of a textual chat, mainly deal with requests for commercial assistance, both on landline and mobile phones. For the purpose of this study, the corpus was extracted by selecting, from a sample of dialogues held over 24 hours, a reduced subset that included requests for explanations from customers. The selection criteria were conceived so as to include all the dialogues where at least one message from the user contained a clearly stated request for explanation. The kind of requests identified in this collection basically reflects the problems typically encountered with a telecom service provider, such as undue or unfamiliar charges in the bill or in the phone credit (about 52% of the overall number of requests in this dataset).

    5The resulting corpus consists of 142 dialogues, with an average of 11 turns per dialogue, and an average length of 9 tokens in customer messages and 38 tokens in the bot messages. Such difference in the message length is due to the way the assistant’s responses are currently structured, in that they usually include detailed information on invoice items or options available, while, on the other hand, customer’s messages are most often quite concise. Also, the relatively high number of turns per dialogue might be explained with the high occurrence in the corpus of repeated or rephrased messages, both by the chatbot and by the customer, due to recurring misunderstandings on both sides.

    Figure 1: Excerpts from the annotated dataset. Annotation of errors is highlighted in red, that of customer’s emotions in blue and the agent’s empa-thy in orange

    (a)

    Image 100002010000003B00000037AE80D84196A55216.png

    Vuoi vedere il dettaglio della fattura?

    (Do you want to see the invoice details?)

    Image 10000201000000310000003067F1080C5A90B0F5.png

    La fattura di gennaio 2020 non arrivata

    (I havent received the invoice of January 2020 yet)

    Ignoring question/feedback - Neutral

    Ecco il dettaglio delle voci di spesa presenti

    nella fattura InvoiceNumber del mese di gennaio

    per la linea PhoneNumber:InvoiceDetails.

    Image 100002010000003B00000037AE80D84196A55216.png

    Vuoi avere il dettaglio di una specifica voce di spesa

    presente in fattura?

    (Here is the detail of the items on the invoice InvoiceNumber

    of January for the PhoneNumber:InvoiceDetails.

    Do you want the detail of a specific item in the invoice?)

    Straight wrong response

    Image 10000201000000310000003067F1080C5A90B0F5.png

    Non mi interessa questa fattura, mi serve gennaio 2020

    (I dont need this invoice, I need the one of January 2020)

    Repetition - Somewhat frustrated

    Image 100002010000003B00000037AE80D84196A55216.png

    Vuoi cambiare argomento?

    (Do you want to change topic?)

    Topic change

    Image 10000201000000310000003067F1080C5A90B0F5.png

    Ciao

    (Whatever)

    Non-cooperativity - Somewhat frustrated

    (b)

    Image 100002010000003B00000037AE80D84196A55216.png

    Scusami, non ho capito bene la tua richiesta...

    Puoi ripeterla in modo pi semplice?

    (Im sorry, I didnt get your question...

    Could you rephrase it in a simpler way?)

    Image 10000201000000310000003067F1080C5A90B0F5.png

    No non hai capito niente

    (No you got it all wrong)

    Non cooperativity - Somewhat angry

    Image 100002010000003B00000037AE80D84196A55216.png

    Mi spiace non aver compreso correttamente.

    Ho la possibilit di farti chattare con un operatore.

    umano pi esperto. Vuoi?

    (Im sorry I havent understood that correctly.

    I can put you in contact with a human representative.

    Is this what you want?)

    Empathy

    6As a matter of fact, the presence of such phenomena in the corpus, along with the overall goals set forth for the development of the NLG module in this project, led us to the design of an annotation process that involved different dimensions, such as errors in conversation and emotions. By error, in this context, we mean any event that might have a negative impact on the flow of the interaction, and more generally on its quality, potentially resulting in breakdowns (i.e. whenever one party leaves the conversation without completing the given task (Martinovsky and Traum 2003)). The error tagset used in this corpus is partially inspired by three of the popular Gricean maxims, i.e. those of Quantity, Relation and Manner (Grice 1989) (each one including further sub-types, not described here) , and it has been conceived so as to include error categories that may apply to both conversation parties. The second dimension, instead, is meant to include, on the one hand, customers’ emotions (as perceived in their messages), and, on the other hand, the chatbot’s empathic responses (if any). In particular, as regards customers’ emotions, besides two generic labels for neutral and positive messages, we mostly focused on negative emotions, especially anger and frustration, also introducing for these ones two finer-grained labels that define their lower or higher intensity. While a full description of the annotation scheme is beyond the scope of this paper, Figure 1 shows two brief examples of how we applied this scheme to the sample dataset2. An overview of the scheme with a discussion on the main findings and annotation issues can be found in Sanguinetti et al. (2020).

    7Due to privacy concerns and the related anonymization issues that may arise (as further discussed in Section 4), the corpus cannot yet be publicly released. However, in an attempt to provide a qualitative analysis of the annotated data, we collected some basic statistics on the distribution of errors and emotions labeled in this sample set. Overall, we report an amount of 326 errors (about 21% of the total number of turns) from both parties; among them, the error class that includes violations of the maxim of Relevance is by far the most frequent one (65% of the errors). Such violations may take different forms, also depending on whether they come from the customer or the chatbot. As regards the customer, errors of such kind typically take place when the user does not take into account the previous message from the chatbot, thus providing irrelevant responses that do not allow to move forward with the conversation and make any progress; these cases cover approximately 21% of customers’ errors. On the chatbot side, the most frequent error type is represented by those cases in which the agent misinterprets a previous customer’s message and proposes to move on to another topic rather than providing a proper response (30% of cases). As for the second annotation dimension, i.e. the one regarding customers’ emotions, most of the messages have a neutral tone (about 86% of user turns), but, among non-neutral messages, the two main negative emotions defined in this scheme, namely anger and frustration, are the ones most frequently encountered in user messages (both with a frequency of 41%), while the cases of messages with a positive emotion constitute less than 1%, and usually translate into some form of gratitude, appreciation, or simple politeness.

    8All these dimensions are functional to a further development of the NLG module, in that they provide, through different perspectives, useful signals of how, and at which point in the conversation, the template response currently used by the chatbot might be improved using the NLG module. Broadly speaking, framing the error taxonomy within the Grice’s cooperative principle provides a useful support for the generation module to understand, in case an error is reported, how to structure the chatbot response so as to improve the interaction quality in terms of informativeness and relevance (as also discussed in Section 3).

    3. Balancing information sources in NLG for DS

    9In this Section, we illustrate a DS architecture that explicitly accounts for the L-DC-UM information channels. In particular, we point out that DC and UM channels can be managed by using standard NLG methods.

    Figure 2: A dialogue system architecture accounting for L-DC-UM channels

    Image 10000000000003B00000021357DAB7C9B9A8AAF2.jpg

    10A commonly adopted architecture for NLG in data-to-text systems is a pipeline composed of four modules: data analyzer, text planner, sentence planner and surface realizer (Reiter 2007; Pauws et al. 2019). Each module tackles a specific issue: (1) the data analyzer determines what can be said, i.e. a domain-specific analysis of input data; (2) the text planner determines what to say, i.e. which information will be communicated; (3) the sentence planner determines how to communicate, with particular attention to the design of the features related to the given content and language (e.g. lexical choices, verb tense, etc.); (4) the surface realizer produces the sentences by using the results of the previous modules and considering language-specific constraints as well. Note that by definition NLG does not account for linguistic input (that is, L-channel), all the modules account for the context of the communication. In other words, data analysis and text planning explicitly process the information about the input data (the DC-channel), and text planning and sentence planning process the information about the audience (the UM-channel). Moreover, by using the nomenclature defined in (Reiter and Dale 2000), the specific task of content selection decides what to say, that is the atomic nucleus of information that will be communicated.

    11In our project, we adopt a complete NLG architecture in the design of the DS (Figure 2). In Figure 2, we show the contributions of the L-DC-UM channels in the interaction flow. It is worth noting that we assigned the content selection task to the DM module rather than to the text planning of the NLG module. Indeed, the content selection task is crucially the point where all the three information channels need to be merged in order to decide the content of the DS answer to the user question.

    12In order to understand the contribution of the three information channels to the final message construction, we describe below the main steps of the module design using the following customer’s message, retrieved from the corpus, as an example:
    Scusami ma vorrei sapere come mai mi vengono fatti alcuni addebiti?. (“Excuse me, I’d like to know why I’m charged certain fees?")

    13Here, the customer requests for an explanation about some (unspecified) charges on her/his bill, making the whole message not informative enough. In this case, the DS can deduce from the L-channel only a generic request of information on transactions. However, using the architecture shown in Figure 2, a more informative answer can be produced considering the UM-channel and the DC-channel.

    14As a working hypothesis, we assume that the user model consists uniquely in the age of the user. By assuming that the user is 18 years old, we can say that the DS should use an informal register, i.e. the Italian second person singular (tu) rather than the more formal third person singular (lei). It is worth noting that the current accounting of the user model is too simple and there is room for improvement both in the formalization of the model, and in the effect of the user model on the generated text. Taking into account the classification of the user model acquisition given by (Reiter, Sripada, and Williams 2003), it is interesting to note that the dialogic nature of the system allow for the possibility to explicitly ask users about their knowledge and preferences on the specific domain.

    15Moreover, we assume that the DC-channel consists of all the transactions of the last 7 months, for example: T1, with an amount of 9.99€ (M1-M7); T2 with an amount of 2€ (M5-M7, appearing twice in M7); and T3 with an amount of 1.59€ (M7) (see Table 1).

    Table 1: A possible transactions history

    M1

    M2

    M3

    M4

    M5

    M6

    M7

    T1

    9.99

    9.99

    9.99

    9.99

    9.99

    9.99

    9.99

    T2

    0

    0

    0

    0

    2

    2

    2, 2

    T3

    0

    0

    0

    0

    0

    0

    1.59

    16Looking at the data in Table 1, different forms of automatic reasoning could be applied in order to evaluate the relevance of each singular transaction of the user. At this stage of the project, we aim to adapt the theory of importance-effect from (Biran and McKeown 2017) to our specific domain, where the relevant information is in the form of relational database entries. The idea is to consider the time evolution of a specific transaction category, giving more emphasis to information contents that can be classified as exceptional evidences. Informally, we can say that the transactions T2 and T3 have a more irregular evolution in time with respect to T1, therefore they should be mentioned with more emphasis in the final message.

    17The current implementation of the DS is based on a trivial NLU (regular-expressions), a symbolic sentence planner and realizer (for Italian) (Anselma and Mazzei 2018; Mazzei, Battaglino, and Bosco 2016). By considering all the three L-UM-DC channels, the answer generated by the DS is:
    Il totale degli addebiti è €15,58. Hai pagato €4,00 (2\times€2,00) per l’Offerta Base Mobile e €1,59 per l’Opzione ChiChiama e RiChiama. Infine, hai pagato il rinnovo dell’offerta 20 GB mobile. (“The total charge is €15.58. You have been charged €4.00(2\times€2.00) for the Mobile Base Offer and €1.59 for the Who’sCalling and CallNow options. Finally, you have been charged for the renewal of the 20 GB mobile offer.”)

    4. Conclusion and Future Work

    18In this paper we have discussed the main features of the design of a DS system for telco customer care. In particular, we outlined the peculiarities of this domain, describing the construction of a specifically-designed dialogue corpus and discussing a possible integration of standard DS and NLG architectures in order to manage these peculiarities. This is an ongoing project and we are considering various enhancements: (1) we will integrate emoji prediction capabilities into the proposed architecture in order to allow the DS to automatically attach an appropriate emoji at the end of the generated response, relying on previous work for Italian (Ronzano et al. 2018); we would also take into account the current user emotions, while generating an appropriate emoji – it may be the case that an emoji that is adequate when the conversation is characterized by a neutral tone, suddenly becomes inappropriate if the user is frustrated or angry (Pamungkas 2019; Cercas Curry and Rieser 2019); (2) we would like to enhance the system so as to adapt the generated responses to other aspects of the users, such as their mental models, levels of domain expertise, and personality traits; (3) we want to evaluate the DS following the user-based comparative schema adopted in (Demberg, Winterboer, and Moore 2011).

    19Finally, we add some closing remarks on the corpus availability and its anonymization. The publication of a dataset of conversations between customers and a company virtual assistant is a great opportunity for the company and for its surrounding communities of academics, designers, and developers. However, it entails a number of obstacles to overcome. Rules and laws by regulating bodies must be strictly followed – see, for example, the GDPR regulation3. This means, first of all, including within the to-be-published dataset only those conversations made by customers who have given their consent to this type of treatment of their data. Moreover, it is mandatory to obscure both personal and sensitive customer data. Such obfuscation activities are particularly difficult in the world of chatbots, where customers are free to input unrestricted text in the conversations. Regular expressions can be used in order to recognize the pieces of data to be obscured, such as email addresses, telephone numbers, social security numbers, bank account identifiers, dates of birth, etc. More sophisticated techniques needed be adopted to identify and obscure, within the text entered by customers, names, surnames, home and work addresses. Even more complex and open is the problem of anonymizing sensitive customer data. For example, consider the case of a disabled customer who reveals his/her sanitary condition to the virtual assistant, in order to obtain a legitimate better treatment from the company: the text revealing the health condition of the customer must be obscured. Other relevant sensitive data include racial or ethnic origins, religious or philosophical beliefs, political opinions, etc. Some of these techniques, used for identifying certain types of data to be obscured, have a certain degree of precision that may even be far, given the current state of the art, from what a trained human analyst could do. Therefore, it is also necessary to consider the need for the dataset being published to be reviewed and edited by specialized personnel before the actual publication. With this in mind, the techniques of data recognition mentioned above - regular expressions, Named Entity Recognition, etc. - could also be exploited to develop tools that can speed up the task of completing and verifying the accurate anonymization of the dataset.

    Bibliographie

    Des DOI sont automatiquement ajoutés aux références bibliographiques par Bilbo, l’outil d’annotation bibliographique d’OpenEdition. Ces références bibliographiques peuvent être téléchargées dans les formats APA, Chicago et MLA.

    Format

    • APA
    • Chicago
    • MLA
    Proceedings of the 11th International Conference on Natural Language Generation. (2018). Presented at the Proceedings of the 11th International Conference on Natural Language Generation. https://doi.org/10.18653/v1/w18-65
    Biran, O., & McKeown, K. (2017). Human-Centric Justification of Machine Learning Predictions. Presented at the Twenty-Sixth International Joint Conference on Artificial Intelligence. https://doi.org/10.24963/ijcai.2017/202
    Bobrow, D. G., Kaplan, R. M., Kay, M., Norman, D. A., Thompson, H., & Winograd, T. (1977). GUS, a frame-driven dialog system. Elsevier BV. https://doi.org/10.1016/0004-3702(77)90018-2
    Cercas Curry, A., & Rieser, V. (2019). A Crowd-based Evaluation of Abuse Response Strategies in Conversational Agents. Presented at the Proceedings of the 20th Annual SIGdial Meeting on Discourse and Dialogue. https://doi.org/10.18653/v1/w19-5942
    Demberg, V., Winterboer, A., & Moore, J. D. (2011). A Strategy for Information Presentation in Spoken Dialog Systems. MIT Press - Journals. https://doi.org/10.1162/coli_a_00064
    Gatt, A., & Krahmer, E. (2018). Survey of the State of the Art in Natural Language Generation: Core tasks, applications and evaluation. AI Access Foundation. https://doi.org/10.1613/jair.5477
    Proceedings of the Eleventh European Workshop on Natural Language Generation - ENLG ’07. (2007). Presented at the the Eleventh European Workshop. https://doi.org/10.3115/1610163
    REITER, E., & DALE, R. (1997). Building applied natural language generation systems. Cambridge University Press (CUP). https://doi.org/10.1017/s1351324997001502
    Traum, D. R., & Larsson, S. (2003). The Information State Approach to Dialogue Management. Springer Netherlands. https://doi.org/10.1007/978-94-010-0019-2_15
    Deemter, K. van, Theune, M., & Krahmer, E. (2005). Real versus Template-Based Natural Language Generation: A False Opposition?. MIT Press - Journals. https://doi.org/10.1162/0891201053630291
    “Proceedings of the 11th International Conference on Natural Language Generation”. []. Association for Computational Linguistics, 2018. doi:10.18653/v1/w18-65.
    Biran, Or, and Kathleen McKeown. “Human-Centric Justification of Machine Learning Predictions”. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence Organization, August 2017. doi:10.24963/ijcai.2017/202.
    Bobrow, Daniel G., Ronald M. Kaplan, Martin Kay, Donald A. Norman, Henry Thompson, and Terry Winograd. “GUS, a Frame-Driven Dialog System”. Artificial Intelligence. Elsevier BV, April 1977. doi:10.1016/0004-3702(77)90018-2.
    Cercas Curry, Amanda, and Verena Rieser. “A Crowd-Based Evaluation of Abuse Response Strategies in Conversational Agents”. Proceedings of the 20th Annual SIGdial Meeting on Discourse and Dialogue. Association for Computational Linguistics, 2019. doi:10.18653/v1/w19-5942.
    Demberg, Vera, Andi Winterboer, and Johanna D. Moore. “A Strategy for Information Presentation in Spoken Dialog Systems”. Computational Linguistics. MIT Press - Journals, September 2011. doi:10.1162/coli_a_00064.
    Gatt, Albert, and Emiel Krahmer. “Survey of the State of the Art in Natural Language Generation: Core Tasks, Applications and Evaluation”. Journal of Artificial Intelligence Research. AI Access Foundation, January 27, 2018. doi:10.1613/jair.5477.
    “Proceedings of the Eleventh European Workshop on Natural Language Generation - ENLG ’07”. []. Association for Computational Linguistics, 2007. doi:10.3115/1610163.
    REITER, EHUD, and ROBERT DALE. “Building Applied Natural Language Generation Systems”. Natural Language Engineering. Cambridge University Press (CUP), March 1997. doi:10.1017/s1351324997001502.
    Traum, David R., and Staffan Larsson. “The Information State Approach to Dialogue Management”. Text, Speech and Language Technology. Springer Netherlands, 2003. doi:10.1007/978-94-010-0019-2_15.
    Deemter, Kees van, Mariët Theune, and Emiel Krahmer. “Real Versus Template-Based Natural Language Generation: A False Opposition?”. Computational Linguistics. MIT Press - Journals, March 2005. doi:10.1162/0891201053630291.
    Proceedings of the 11th International Conference on Natural Language Generation. [], Association for Computational Linguistics, 2018. Crossref, https://doi.org/10.18653/v1/w18-65.
    Biran, Or, and Kathleen McKeown. “Human-Centric Justification of Machine Learning Predictions”. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, International Joint Conferences on Artificial Intelligence Organization, Aug. 2017. Crossref, https://doi.org/10.24963/ijcai.2017/202.
    Bobrow, Daniel G., et al. “GUS, a Frame-Driven Dialog System”. Artificial Intelligence, vols. 8, nos. 2, Elsevier BV, Apr. 1977, pp. 155-73. Crossref, https://doi.org/10.1016/0004-3702(77)90018-2.
    Cercas Curry, Amanda, and Verena Rieser. “A Crowd-Based Evaluation of Abuse Response Strategies in Conversational Agents”. Proceedings of the 20th Annual SIGdial Meeting on Discourse and Dialogue, Association for Computational Linguistics, 2019. Crossref, https://doi.org/10.18653/v1/w19-5942.
    Demberg, Vera, et al. “A Strategy for Information Presentation in Spoken Dialog Systems”. Computational Linguistics, vols. 37, nos. 3, MIT Press - Journals, Sept. 2011, pp. 489-3. Crossref, https://doi.org/10.1162/coli_a_00064.
    Gatt, Albert, and Emiel Krahmer. “Survey of the State of the Art in Natural Language Generation: Core Tasks, Applications and Evaluation”. Journal of Artificial Intelligence Research, vols. 61, AI Access Foundation, 27 Jan. 2018, pp. 65-170. Crossref, https://doi.org/10.1613/jair.5477.
    Proceedings of the Eleventh European Workshop on Natural Language Generation - ENLG ’07. [], Association for Computational Linguistics, 2007. Crossref, https://doi.org/10.3115/1610163.
    REITER, EHUD, and ROBERT DALE. “Building Applied Natural Language Generation Systems”. Natural Language Engineering, vols. 3, no. 1, Cambridge University Press (CUP), Mar. 1997, pp. 57-87. Crossref, https://doi.org/10.1017/s1351324997001502.
    Traum, David R., and Staffan Larsson. “The Information State Approach to Dialogue Management”. Text, Speech and Language Technology, Springer Netherlands, 2003, pp. 325-53. Crossref, https://doi.org/10.1007/978-94-010-0019-2_15.
    Deemter, Kees van, et al. “Real Versus Template-Based Natural Language Generation: A False Opposition?”. Computational Linguistics, vols. 31, no. 1, MIT Press - Journals, Mar. 2005, pp. 15-24. Crossref, https://doi.org/10.1162/0891201053630291.

    Cette bibliographie a été enrichie de toutes les références bibliographiques automatiquement générées par Bilbo en utilisant Crossref.

    Luca Anselma and Alessandro Mazzei. 2018. “Designing and Testing the Messages Produced by a Virtual Dietitian.” In Proceedings of the 11th International Conference on Natural Language Generation, Tilburg University, the Netherlands, November 5-8, 2018, 244–53.

    10.18653/v1/W18-65 :

    Or Biran, and Kathleen McKeown. 2017. “Human-Centric Justification of Machine Learning Predictions.” In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI-17, 1461–7. https://doi.org/10.24963/ijcai.2017/202.

    10.24963/ijcai.2017/202 :

    Daniel G. Bobrow, Ronald M. Kaplan, Martin Kay, Donald A. Norman, Henry Thompson, and Terry Winograd. 1977. “GUS, a Frame-Driven Dialog System.” Artif. Intell. 8 (2): 155–73. https://doi.org/10.1016/0004-3702(77)90018-2.

    10.1016/0004-3702(77)90018-2 :

    Amanda Cercas Curry, and Verena Rieser. 2019. “A Crowd-Based Evaluation of Abuse Response Strategies in Conversational Agents.” In Proceedings of the 20th Annual Sigdial Meeting on Discourse and Dialogue, 361–66. Stockholm, Sweden: Association for Computational Linguistics. https://doi.org/10.18653/v1/W19-5942.

    10.18653/v1/W19-5942 :

    Vera Demberg, Andi Winterboer, and Johanna D. Moore. 2011. “A Strategy for Information Presentation in Spoken Dialog Systems.” Computational Linguistics 37 (3): 489–539. https://doi.org/10.1162/COLI_a_00064.

    10.1162/COLI_a_00064 :

    Albert Gatt, and Emiel Krahmer. 2018. “Survey of the State of the Art in Natural Language Generation: Core Tasks, Applications and Evaluation.” J. Artif. Intell. Res. 61: 65–170. https://doi.org/10.1613/jair.5477.

    10.1613/jair.5477 :

    Google. 2020. “Dialogflow Documentation.” https://dialogflow.com.

    Paul Grice. 1989. Studies in the Way of Words. Cambridge, Massachussets: Harvard University Press.

    Bilyana Martinovsky, and David Traum. 2003. “The Error Is the Clue: Breakdown in Human-Machine Interaction.” In In Proceedings of the Isca Workshop on Error Handling in Dialogue Systems.

    Bilyana Mazzei, Cristina Battaglino, and Cristina Bosco. 2016. “SimpleNLG-IT: adapting SimpleNLG to Italian.” In Proceedings of the 9th International Natural Language Generation Conference, 184–92. Edinburgh, UK: Association for Computational Linguistics.

    Michael McTear, Zoraida Callejas, and David Griol. 2016. The Conversational Interface: Talking to Smart Devices. 1st ed. Springer Publishing Company, Incorporated.

    Endang Wahyu Pamungkas. 2019. “Emotionally-Aware Chatbots: A Survey.” CoRR abs/1906.09774. http://arxiv.org/abs/1906.09774.

    Steffen Pauws, Albert Gatt, Emiel Krahmer, and Ehud Reiter. 2019. “Making Effective Use of Healthcare Data Using Data-to-Text Technology.” In Data Science for Healthcare, 119–45. Springer.

    Ehud Reiter. 2007. “An Architecture for Data-to-Text Systems.” In Proc. Of the 11th European Workshop on Natural Language Generation, 97–104. ENLG ’07. Stroudsburg, PA, USA: Association for Computational Linguistics.

    10.3115/1610163 :

    Ehud Reiter and Robert Dale. 2000. Building Natural Language Generation Systems. New York, NY, USA: Cambridge University Press.

    10.1017/S1351324997001502 :

    Ehud Reiter Somayajulu Sripada, and Sandra Williams. 2003. “Acquiring and Using Limited User Models in NLG.” In Proceedings of the 9th European Workshop on Natural Language Generation (ENLG-2003) at EACL 2003.

    Anna Rohrbach, Lisa Anne Hendricks, Kaylee Burns, Trevor Darrell, and Kate Saenko. 2018. “Object Hallucination in Image Captioning.” In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, 4035–45. Brussels, Belgium: Association for Computational Linguistics.

    Francesco Ronzano, Francesco Barbieri, Endang Wahyu Pamungkas, Viviana Patti, and Francesca Chiusaroli. 2018. “Overview of the EVALITA 2018 Italian Emoji Prediction (ITAMoji) Task.” In Proceedings of the Sixth Evaluation Campaign of Natural Language Processing and Speech Tools for Italian. Final Workshop (EVALITA 2018). Vol. 2263. CEUR Workshop Proceedings. CEUR-WS.org. http://ceur-ws.org/Vol-2263/paper004.pdf.

    David Traum and Staffan Larsson. 2003. “The Information State Approach to Dialogue Management.” In Current and New Directions in Discourse and Dialogue, 325–53. Springer. https://link.springer.com/chapter/10.1007/978-94-010-0019-2_15.

    10.1007/978-94-010-0019-2_15 :

    Kees Van Deemter, Emiel Krahmer, and Mariët Theune. 2005. “Real Versus Template-Based Natural Language Generation: A False Opposition?” Comput. Linguist. 31 (1): 15–24. https://doi.org/10.1162/0891201053630291.

    10.1162/0891201053630291 :

    Notes de bas de page

    1 The well-known problem of hallucinations in neural networks deters their use in real-world NLG (Rohrbach et al. 2018).

    2 For further details on the scheme and the definition of all tags, the annotation guidelines are available in this document: https://cutt.ly/cdMcnyM

    3 https://eur-lex.europa.eu/eli/reg/2016/679/oj

    Auteurs

    • Mirko Di Lascio

      Dipartimento di Informatica, Università degli Studi di Torino, Italy – mirko.dilascio@unito.it

    • Manuela Sanguinetti

      Dipartimento di Informatica, Università degli Studi di Torino, Italy - Dipartimento di Matematica e Informatica, Università degli Studi di Cagliari, Italy – manuela.sanguinetti@unito.it

    • Luca Anselma

      Dipartimento di Informatica, Università degli Studi di Torino, Italy – luca.anselma@unito.it

    • Dario Mana

      TIM, Torino, Italy – dario.mana@telecomitalia.it

    • Alessandro Mazzei

      Dipartimento di Informatica, Università degli Studi di Torino, Italy – alessandro.mazzei@unito.it

    • Viviana Patti

      Dipartimento di Informatica, Università degli Studi di Torino, Italy – viviana.patti@unito.it

    • Rossana Simeoni

      TIM, Torino, Italy – rossana.simeoni@telecomitalia.it

    Précédent Suivant
    Table des matières

    Le texte seul est utilisable sous licence Licence OpenEdition Books. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

    Voir plus de livres
    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    3-4 December 2015, Trento

    Cristina Bosco, Sara Tonelli et Fabio Massimo Zanzotto (dir.)

    2015

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    5-6 December 2016, Napoli

    Anna Corazza, Simonetta Montemagni et Giovanni Semeraro (dir.)

    2016

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 7 December 2016, Naples

    Pierpaolo Basile, Franco Cutugno, Malvina Nissim et al. (dir.)

    2016

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    11-12 December 2017, Rome

    Roberto Basili, Malvina Nissim et Giorgio Satta (dir.)

    2017

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    10-12 December 2018, Torino

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian

    EVALITA Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 12-13 December 2018, Naples

    Tommaso Caselli, Nicole Novielli, Viviana Patti et al. (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian Final Workshop

    Valerio Basile, Danilo Croce, Maria Maro et al. (dir.)

    2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Bologna, Italy, March 1-3, 2021

    Felice Dell'Orletta, Johanna Monti et Fabio Tamburini (dir.)

    2020

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Milan, Italy, 26-28 January, 2022

    Elisabetta Fersini, Marco Passarotti et Viviana Patti (dir.)

    2022

    Voir plus de livres
    1 / 9
    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    Proceedings of the Second Italian Conference on Computational Linguistics CLiC-it 2015

    3-4 December 2015, Trento

    Cristina Bosco, Sara Tonelli et Fabio Massimo Zanzotto (dir.)

    2015

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    Proceedings of the Third Italian Conference on Computational Linguistics CLiC-it 2016

    5-6 December 2016, Napoli

    Anna Corazza, Simonetta Montemagni et Giovanni Semeraro (dir.)

    2016

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    EVALITA. Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 7 December 2016, Naples

    Pierpaolo Basile, Franco Cutugno, Malvina Nissim et al. (dir.)

    2016

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    Proceedings of the Fourth Italian Conference on Computational Linguistics CLiC-it 2017

    11-12 December 2017, Rome

    Roberto Basili, Malvina Nissim et Giorgio Satta (dir.)

    2017

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    Proceedings of the Fifth Italian Conference on Computational Linguistics CLiC-it 2018

    10-12 December 2018, Torino

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian

    EVALITA Evaluation of NLP and Speech Tools for Italian

    Proceedings of the Final Workshop 12-13 December 2018, Naples

    Tommaso Caselli, Nicole Novielli, Viviana Patti et al. (dir.)

    2018

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    EVALITA Evaluation of NLP and Speech Tools for Italian - December 17th, 2020

    Proceedings of the Seventh Evaluation Campaign of Natural Language Processing and Speech Tools for Italian Final Workshop

    Valerio Basile, Danilo Croce, Maria Maro et al. (dir.)

    2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Bologna, Italy, March 1-3, 2021

    Felice Dell'Orletta, Johanna Monti et Fabio Tamburini (dir.)

    2020

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Proceedings of the Eighth Italian Conference on Computational Linguistics CliC-it 2021

    Milan, Italy, 26-28 January, 2022

    Elisabetta Fersini, Marco Passarotti et Viviana Patti (dir.)

    2022

    Voir plus de chapitres

    Experimenting the use of catenae in Phrase-Based SMT

    Manuela Sanguinetti

    Query in linguaggio naturale per il dominio della dieta mediterranea

    Luca Anselma, Dario Ferrero et Alessandro Mazzei

    Generare messaggi persuasivi per una dieta salutare

    Alessandro Mazzei

    Building a computational lexicon by using SQL

    Alessandro Mazzei

    HaSpeeDe 2 @ EVALITA2020: Overview of the EVALITA 2020 Hate Speech Detection Task

    Manuela Sanguinetti, Gloria Comandini, Elisa di Nuovo et al.

    Raising Interest and Collecting Suggestions on the EVALITA Evaluation Campaign

    Rachele Sprugnoli, Viviana Patti et Franco Cutugno

    Neural Surface Realization for Italian

    Valerio Basile et Alessandro Mazzei

    Evalita 2018: Overview on the 6th Evaluation Campaign of Natural Language Processing and Speech Tools for Italian

    Tommaso Caselli, Viviana Patti, Nicole Novielli et al.

    Preface

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini

    Dealing with Italian Adjectives in Noun Phrase: a study oriented to Natural Language Generation

    Giorgia Conte, Cristina Bosco et Alessandro Mazzei

    Personal-ITY: A Novel YouTube-based Corpus for Personality Prediction in Italian

    Elisa Bassignana, Malvina Nissim et Viviana Patti

    Voir plus de chapitres
    1 / 11

    Experimenting the use of catenae in Phrase-Based SMT

    Manuela Sanguinetti

    Query in linguaggio naturale per il dominio della dieta mediterranea

    Luca Anselma, Dario Ferrero et Alessandro Mazzei

    Generare messaggi persuasivi per una dieta salutare

    Alessandro Mazzei

    Building a computational lexicon by using SQL

    Alessandro Mazzei

    HaSpeeDe 2 @ EVALITA2020: Overview of the EVALITA 2020 Hate Speech Detection Task

    Manuela Sanguinetti, Gloria Comandini, Elisa di Nuovo et al.

    Raising Interest and Collecting Suggestions on the EVALITA Evaluation Campaign

    Rachele Sprugnoli, Viviana Patti et Franco Cutugno

    Neural Surface Realization for Italian

    Valerio Basile et Alessandro Mazzei

    Evalita 2018: Overview on the 6th Evaluation Campaign of Natural Language Processing and Speech Tools for Italian

    Tommaso Caselli, Viviana Patti, Nicole Novielli et al.

    Preface

    Elena Cabrio, Alessandro Mazzei et Fabio Tamburini

    Dealing with Italian Adjectives in Noun Phrase: a study oriented to Natural Language Generation

    Giorgia Conte, Cristina Bosco et Alessandro Mazzei

    Personal-ITY: A Novel YouTube-based Corpus for Personality Prediction in Italian

    Elisa Bassignana, Malvina Nissim et Viviana Patti

    Accès ouvert

    Accès ouvert

    ePub

    PDF

    PDF du chapitre

    1 The well-known problem of hallucinations in neural networks deters their use in real-world NLG (Rohrbach et al. 2018).

    2 For further details on the scheme and the definition of all tags, the annotation guidelines are available in this document: https://cutt.ly/cdMcnyM

    3 https://eur-lex.europa.eu/eli/reg/2016/679/oj

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    X Facebook Email

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Ce livre est cité par

    • Miaschi, Alessio. Alzetta, Chiara. Brunato, Dominique. Dell’Orletta, Felice. Venturi, Giulia. (2023) Testing the Effectiveness of the Diagnostic Probing Paradigm on Italian Treebanks. Information, 14. DOI: 10.3390/info14030144
    • Miaschi, Alessio. Sarti, Gabriele. Brunato, Dominique. Dell’Orletta, Felice. Venturi, Giulia. (2022) Probing Linguistic Knowledge in Italian Neural Language Models across Language Varieties. Italian Journal of Computational Linguistics, 8. DOI: 10.4000/ijcol.965
    • Saga, Satoshi. Shirakawa, Kanta. (2025) Investigating the Effect of Pseudo-Haptics on Perceptions Toward Onomatopoeia Text During Finger-Point Tracing. Multimodal Technologies and Interaction, 9. DOI: 10.3390/mti9100100

    Ce chapitre est cité par

    • Mazzei, Alessandro. Anselma, Luca. Sanguinetti, Manuela. Rapp, Amon. Mana, Dario. Hossain, Md. Murad. Patti, Viviana. Simeoni, Rossana. Longo, Lucia. (2022) Anticipating User Intentions in Customer Care Dialogue Systems. IEEE Transactions on Human-Machine Systems, 52. DOI: 10.1109/THMS.2022.3184400

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Vous pouvez vous connecter à votre bibliothèque à l’adresse suivante : https://freemium.openedition.org/oebooks

    Suggérer l’acquisition à votre bibliothèque

    Si vous avez des questions, vous pouvez nous écrire à access[at]openedition.org

    Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020

    Vérifiez si votre bibliothèque a déjà acquis ce livre : authentifiez-vous à OpenEdition Freemium for Books.

    Vous pouvez suggérer à votre bibliothèque d’acquérir un ou plusieurs livres publiés sur OpenEdition Books. N’hésitez pas à lui indiquer nos coordonnées : access[at]openedition.org

    Vous pouvez également nous indiquer, à l’aide du formulaire suivant, les coordonnées de votre bibliothèque afin que nous la contactions pour lui suggérer l’achat de ce livre. Les champs suivis de (*) sont obligatoires.

    Veuillez, s’il vous plaît, remplir tous les champs.

    La syntaxe de l’email est incorrecte.

    Référence numérique du chapitre

    Format

    Di Lascio, M., Sanguinetti, M., Anselma, L., Mana, D., Mazzei, A., Patti, V., & Simeoni, R. (2020). Natural Language Generation in Dialogue Systems for Customer Care. In F. Dell’Orletta, J. Monti, & F. Tamburini (éds.), Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020. Torino: Accademia University Press. https://doi.org/10.4000/books.aaccademia.8450
    Di Lascio, Mirko, Manuela Sanguinetti, Luca Anselma, Dario Mana, Alessandro Mazzei, Viviana Patti, et Rossana Simeoni. « Natural Language Generation in Dialogue Systems for Customer Care ». In Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020, édité par Felice Dell’Orletta, Johanna Monti, et Fabio Tamburini. Torino: Accademia University Press, 2020. doi:10.4000/books.aaccademia.8450.
    Di Lascio, Mirko, et al. « Natural Language Generation in Dialogue Systems for Customer Care ». Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020, édité par Felice Dell’Orletta et al., Accademia University Press, 2020, https://doi.org/10.4000/books.aaccademia.8450.

    Référence numérique du livre

    Format

    Dell’Orletta, F., Monti, J., & Tamburini, F. (éds.). (2020). Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-it 2020. Torino: Accademia University Press. https://doi.org/10.4000/books.aaccademia.8203
    Dell’Orletta, Felice, Johanna Monti, et Fabio Tamburini, éd. Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020. Torino: Accademia University Press, 2020. doi:10.4000/books.aaccademia.8203.
    Dell’Orletta, Felice, et al., éditeurs. Proceedings of the Seventh Italian Conference on Computational Linguistics CLiC-It 2020. Accademia University Press, 2020, https://doi.org/10.4000/books.aaccademia.8203.
    Compatible avec Zotero Zotero

    1 / 3

    Accademia University Press

    Accademia University Press

    • Plan du site
    • Se connecter

    Suivez-nous

    • Facebook
    • Flux RSS

    URL : http://www.aaccademia.it/

    Email : info@aaccademia.it

    Adresse :

    Accademia University Press

    Via Carlo Alberto 55

    I‐10123

    Torino

    Italia

    OpenEdition
    • Candidater à OpenEdition Books
    • Connaître le programme OpenEdition Freemium
    • Commander des livres
    • S’abonner à la lettre d’OpenEdition
    • CGU d’OpenEdition Books
    • Accessibilité : partiellement conforme
    • Données personnelles
    • Gestion des cookies
    • Système de signalement