NoVAGraphS is an ongoing project on educational-oriented dialogue systems, specifically designed to be accessible by visually impaired individuals. The main goal of the project is the design and the realization of conversational interfaces for accessing images with an internal graph structure such as UML and E-R diagrams, functional diagrams, and electrical circuits. At the heart of the NoVAGraphS project lies NoVABOT, a dialogue system for Italian and English based on AIML (Artificial Intelligence Markup Language), an open standard scripting language for dialogue systems. We devised NoVABOT as a web application, implementing a textual dialogue system fully compliant with the Web Content Accessibility Guidelines 2.1, with the aim to allow blind people to use their usual speech-recognition/text2speech interfaces (a screen reader ). The development of the dialogue system was carried out using an open-source library that was extended with new features to enable the manipulation of mathematical formulas and images.
The rapid advancement of Natural Language Processing (NLP) can directly benefit the development of systems dedicated for students. In particular, educational Dialogue Systems (DSs) can play the role of a personal tutor for assisting the learning process [1].
In educational settings, ensuring equal opportunities for all students is important. However, Visually Impaired People (VIP) often face challenges in accessing images and graphical structures—such as tables, trees, UML diagrams, E-R diagrams, and circuits—due to inadequate alt-text and image descriptions. Unlike established technologies like tactile and haptic interfaces, which have known limitations [2], LGOBE ∗Corresponding author.
CEUR
ceur-ws.org
Natural Language Processing/Generation (NLP/G) is a relatively unexplored but promising approach for conveying graphical information efectively. The widespread availability of speech-to-text and text-to-speech functionalities in modern electronic devices highlights the potential of leveraging these technologies to reduce accessibility barriers.
NoVAGraphS1 (Non Visual Access to Graphical Structures) is an ongoing project devoted to the
development of educational DSs for VIP that specifically address the treatment of graphical structures.
The main goals of the project are: (
In [3], we presented both (
In this paper we present the main features of NoVABOT, a new DS specifically designed to fulfill the goals of the NoVAGraphS project considering the experience and the results of [3]. There are five new features that empower NoVABOT with respect to FSAEnDS, that are: • NoVABOT has been realized by accounting three diferent domains (see Section 3.3) related to graphical structures, that are functional diagrams from maths, electrical circuits from physics, Finite State Aautomata from computer science. FSAEnDS considered only the latter. • NoVABOT has been designed for both VIP and non-VIP final users . The idea is to consider the possible users’ impairment as just one feature of user modelling. Our assumption is that NoVABOT can be profitably used by diferent kinds of users that can have (or not have) diferent kinds of impairments. As a consequence, in contrast to FSAEnDS, the NoVABOT interface is a mix of graphics and texts (see Section 3). • In NoVABOT, we implemented a procedure for the automatic creation of interaction scripts related to the graphical structures. On the basis of the FSAEnDS, where the interaction scripts have been entirely created from scratch by humans using the happy-path methodology, we exploit the vectorial nature of graphical structures. The idea is to convert sub-parts of the graphical structures into adjacency pairs of the dialogue script (see Section 3.4). • NoVABOT has been designed for dialogues in Italian (see Section 3), in contrast to FSAEnDS that was designed for English. • We evaluate NoVABOT by using pedagogy of mathematics methodologies. In particular, focusing on the mathematical domain, we collect and analyze the feedback of both (i) 3 VIP students (see Section 4.1), and (ii) 8 specialized teachers on accessible technologies (see Section 4.2). We believe that these two initial evaluations can shed light on the future development of NoVABOT and give a valuable test on the applicability of DSs to the education of VIP.
In the remaining part of the paper, we give a brief report of related work in Section 2, we describe the main features of NoVABOT in Section 3, we report the results of evaluation in Section 4, and finally we close the paper with conclusions and future work in Section 5.
There are three main approaches in accessible technologies literature for accounting the problem of presenting the information enclosed in graphical structures to VIP.
Haptic representations, that are tactile information as vibration, touch, and force feedback, have been proposed in multi-modal systems for communicating graphical information in several domains as, for instance, graphs [5], chemical formulas [6] and function graphs [7].
Sonification , that is practice of representing data through sound, has been studied in diferent domains to explore as, for instance, function graphs [8], elementary geometric shapes [9] and maps [10].
Finally, textual descriptions of images have been studied for representing, for instance, chemical formulas [11], electronic circuits [12] and function graphs [13].
However, all these studies point out two main limitations of such solutions. First, a strong knowledge of the domain and characteristics of non-visual perception is required on the part of the person making the textual description. Second, the cognitive load to understand the textual description of a complex image is very high since the exploration is strictly sequential, without the possibility from the VIP to ask for clarification or repetition of the presented information.
Some academic and commercial works applied NLG to describe to VIP some specific structures, typical of scientific communication. For instance, diferent NLG techniques have been applied to produce descriptions of bar charts [14], and to the problem of communicating mathematical expressions [15, 16, 17].
Finally, to our knowledge, the only works that use a DS for the specific problem of communicating graphical information to VIP are [3] and [18].
In this section, we provide a description of NoVABOT, defining its architecture (Section 3.1), the extension of an open-source library that manages AIML (Section 3.2), the application domains (Section 3.3), a procedure for automatic conversion of a structured image in SVG into AIML (Section 3.4), and, finally, the interface (Section 3.5).
Note that NoVABOT is a web-based and accessible text-based DS that can be accessed by VIP using their own speech technologies, usually a screen reader (a dedicated software that reads every item on the screen) and a keyboard. In this way we can evaluate the impact of the dialogic interaction without altering the VIP final physical interaction. Indeed, the use of text-to-speech technologies not specifically designed for VIP can degrade the user experience of VIP users [15, 17].
NoVABOT3 has been integrated into a web application, developed using a number of standard web library (HTML, JavaScript, Node.js, Bootstrap). The web application follows a typical client-server architecture. The backend, powered by Node.js, manages the business logic, processes user requests, generates responses, and handles conversation logic using AIML. The frontend, which consists of HTML and JavaScript with Bootstrap and EJS, provides an interactive user interface. Bootstrap aids in layout design, while EJS dynamically generates HTML pages based on responses processed by the backend, ensuring a responsive and dynamic user experience. The code of NoVABOT can be found on GitHub4.
The development of NoVABOT was carried out using AIML5, an XML-based language designed for developing DS. More specifically, AIML is a declarative language that follows the pattern-matching paradigm. Rule-based systems, by design, provide consistent and controlled responses, though this comes at the expense of flexibility and scalability.
The AIML-high library6 was initially used to interpret the AIML files. This interpreter could handle several common AIML tags (Table 1). However, we extended the interpreter by adding some new functionalities described below. Managing Multiple Wildcards This new functionality involved correcting an error in the AIMLhighlibrary, where multiple wildcards were not handled correctly. For example, given the pattern: (HELLO * I AM *), which could be triggered by a query like: Hello Mario, I am John. It was expected that the interpreter would assign the first wildcard to the string Mario and the second to the string John. However, the library was erroneously assigning both wildcards to the string Luca, indicating a wrong management of the wildcards. The correction ensured that each wildcard was associated with its respective string.
Formula Tag The extension allowed the inclusion of a <formula> tag within AIML templates, enabling the insertion of a mathematical formula conforming to MathJax syntax. The response generated by the interpreter includes any formula detected in the template within \( \), allowing the MathJax library7 to interpret the formula during the response rendering phase. This capability enhances the NoVABOT utility in academic and technical contexts where precise mathematical notation is crucial. Image Tag To enable the creation of responses containing images, the interpreter has been enhanced to handle the <image> tag. Within this tag, the image name or its path must be specified, depending on how the interpreter’s response is handled on the server side. In addition to specifying the image, the <svgElement> tag has been implemented, which can be used exclusively for SVG images. With this tag, it is possible to modify the style of an image element by specifying the ID of the element to be modified, the style name, and the value to be applied. An example of an AIML rule using these tags is as follows: <image>functional-diagram.svg</image> <svgElement style-name="stroke" style-value="#04ed00">
Mario</svgElement> In this case, activating the rule will return the image name to be rendered and the element to be modified, in addition to the textual response. Specifically, the element with ID Mario will be modified, changing the stroke color to green (stroke:"#04ed00").
The main goal of NoVABOT is the capacity of answering to questions in Italian related to topics of functional diagrams, electrical circuits, finite state automata . To achieve this goal, we focused on designing NoVABOT to understand and process queries in the Italian language.
For the domain of functions, NoVABOT is capable of discussing the fundamentals of set theory, the operations on sets, the types of functions and their properties, and the description of an example function. For the electrical circuits, NoVABOT can assist by answering questions about circuit components, Ohm’s Law, and Kirchhof’s Laws, as well as providing explanations about the composition of a sample circuit. Regarding the FSAs domain, NoVABOT provides explanations about the state and transitions of a specific example.
For classifying and managing the users’ queries, we use dialogue acts, based on ISO 24617-2 [19, 20], to handle diferent type of questions. Specifically, we chose DS:opening, for the initial greetings, and Ta:request, for general questions, such as Parlami degli automi (Tell me about automata) or Descrivimi il circuito elettrico (Describe me the electrical circuit). For specific questions, like Quanti stati ha l’automa? (How many states does the automaton have?) we use Ta:setQuestion. Additionally, we use Ta:propositionalQuestion for questions that require confirmation or refutation.
We managed theoretical questions by creating AIML rules with a predefined answer for the initial occurrence of each specific question and varied responses for subsequent repetitions. To accomplish this, we utilized an AIML variable to track whether the question had been asked before. For repeated questions, we employed the <random> tag, which selects a response randomly from multiple options. This approach was implemented to enhance the system variety when questions are repeated and to provide more comprehensive answers to the user.
For questions regarding specific examples of functional diagrams, electrical circuits and FSA, we built an SVG-to-Text system, which we describe in Section 3.4, to create AIML rules that were subsequently added to the AIML file with the theoretical rules.
The responses related to domain-specific examples were generated automatically. This was achieved by constructing SVG images for each example (function, circuit, automaton), extracting information from them, and using this information to generate responses for building AIML rules. This process employed a Template-based system [21]. Although this technique has limitations, such as the lack of lfexibility and the inability to handle unexpected inputs, it is suitable for our case, since all diagrams have a strict predefined structure.
By using a simple parser we can extract the relevant semantic information, that is: function link-Kevin-Siena link-Luigi-Milano link-Josephine-Milano link-Silvia-Bari domain(name-A)
Kevin,Luigi,Josephine,Silvia codomain(name-B)
Siena,Bologna,Milano,Roma,Bari
With this information, we automatically construct AIML rules. For instance: Pattern: *NOME*DOMINIO* Template Schema: Il nome è domain["name"] Filled Template: Il nome è A (The name is A)
The process involved anticipating questions by following the predefined scenarios and building responses using the Template-based Data-to-Text methodology. This approach ensured that NoVABOT can generate accurate and contextually appropriate responses based on the structured knowledge extracted from the SVG representations. The example shows the specific case of the function; the same approach has been applied to circuits and FSA as well.
Figure 2 shows the user interface for NoVABOT, specifically designed for functional diagrams. Various elements were considered to make it accessible for VIPs.
Firstly, the interface is optimised for screen readers—leveraging the familiarity of a tool that VIP are accustomed to due to their personal voice settings (pitch, speed, etc.), with a large input bar and a wide submit button immediately below it.
Secondly, user responses appear below the submit button, and a table at the bottom shows the conversation history, allowing VIP to match questions with answers easily by using HTML tags to navigate the table.
Lastly, the interface includes an image of the structure being navigated. Though accessible via SVG tags, it is meant mainly for future development. Our goal is to expand this tool for supporting individuals with attention deficit hyperactivity disorder, who can benefit from visual aids and color changes highlighting relevant parts.
Note that the web of NoVABOT interface is fully compliant with the Web Content Accessibility Guidelines 2.18, thus fully accessible to VIP.
Domanda:
Chiedimi qualcosa riguardo agli insiemi e alle funzioni matematiche... L'elemento Marco dell'insieme A è collegato all'elemento 34 dell'insieme B.
Insiemi
Invia A
Luigina
Giovanna
Marco
In order to study the efectiveness of the software for educational purposes, we conducted two diferent experiments. First, we carried out task-based interviews with blind users to assess the software efectiveness for mathematical practice. Second, we administered and analyzed a questionnaire given to primary and secondary school special education teachers and mathematics teachers.
All Right Reserved © NoVAGraphS - Università degli Studi di Torino
We conducted semi-structured, task-based interviews with three users. Two of them were graduated in scientific subjects and experienced users of assistive technologies, and the third was an undergraduate psychology student. All three users were invited to participate in the experiment on a volunteer basis without compensation.
During the interviews, we focused on the use of representations of functions between finite sets. We adapted a task from [22], where diferent examples of injective functions were given, followed by diferent examples of non-injective functions. The interviewed students interacted with NoVABOT to explore the given functions. They were asked to recognize the shared features of the first set of examples and the diferences from the second set, ultimately attempting to define an injective function. The same process was repeated for surjective functions. The original version of the chosen example heavily relied on the visual representation of the given functions using Euler-Venn diagrams.
We observed how the alternative representation provided by NoVABOT allowed our interviewees to explore the diagrams and solve the given problems. Specifically, the interviewees confirmed the utility of analyzing diferent aspects of the function separately by asking NoVABOT specific questions about the number of elements in the domain and codomain or the single connections between elements of the domain and codomain. They then shifted to questions that provided more global information about the function under study allowing its synthetic comprehension.
Nevertheless, the unique representational nature of NoVAGraphS, as compared to diagrammatic graphical representations, implies that, particularly from an educational standpoint, one must be aware that it is not possible to simply “translate” a task represented by a drawing through interaction with the DS. Therefore, instructional design must be sensitive to these diferences. For example, interaction times are significantly diferent, and an “at-a-glance” overview of the diagram under study is still not possible by the user.
We are currently analyzing students’ interactions with the software using the Peircean construct of diagrammatic reasoning [23]. This approach aims to deepen our understanding of how students can use this type of auditory representation to understand and utilize diagrams in problem-solving activities. However, all students experienced some dificulties with NoVABOT, as it was unable to answer many of their questions. A crucial aspect of the users’ experience regarded the robustness of the natural language understanding (NLU) module in NoVABOT. Indeed, in some cases, NoVABOT was unable to answer some questions posed by VIP. We are currently working on an integration layer with a Large Language Model (LLM) to increase the flexibility of the NLU module.
We also allowed both special education teachers and mathematics teachers from diferent school levels to interact with the software to assess its usefulness in teaching practices with real or hypothetical visually impaired students.
Eight teachers were invited to participate in the experiment on a volunteer basis without compensation.
Each teacher stated that they had experience with at least one visually impaired student. The invitation
letter in Italian can be found in Appendix. The teachers were required to interact with NoVABOT for
about a dozen minutes and then complete the User Experience Questionnaire (UEQ) [24]. The UEQ
calculates six scales: (
UEQ Scale Attractiveness Perspicuity Eficiency Dependability Stimulation Novelty
Mean
Variance
Two out of the eight teachers were not included in the analysis of the results because their responses were deemed suspicious (i.e., a problematic data pattern like difering opinions across various scales, or a tendency to always choose the middle category). Table 2 shows the mean and variance results for each scale across the six teachers. Values between −0.8 and 0.8 represent a neutral evaluation, values > 0.8 represent a positive evaluation and values < −0.8 represent a negative evaluation. Given the scale range from −3 (horribly bad) to +3 (extremely good) of the UEQ, the observed values are all positive, indicating that NoVABOT performs well in the assessed areas. In particular, the results indicate a generally positive evaluation across most scales, with means above the neutral threshold of 0.8 for Perspicuity, Eficiency, and Dependability. The high variance observed in the results can be attributed to the relatively small number of users.
Given these results, we can conclude, despite the small sample size of teachers who participated in the study, that NoVABOT is expected to be user-friendly, secure, predictable, and efective in its role as a personalized educational DS. However, in order to assess the robustness of this encouraging results, we plan to repeat the evaluation with a larger number of experimenters.
This paper reports the study, design, and implementation of NoVABOT, a DS to help VIP to access to graphical structures as functional sets, circuits and FSA. We discussed the system architecture, its interface, and the improvements we made to an open source AIML interpretation library. We proposed a strategy to automatically generating textual descriptions of functional diagrams, electrical circuits and FSA starting from their SVG representation by using AIML rules. Additionally, we conducted two evaluations on the specific domain of functional diagrams: one involving blind students and another with teachers who work daily with VIP. Despite the small number of individuals involved in the experiment, on the one hand, these experiments revealed how VIP benefit from using NoVABOT and how the alternative dialogical representation provided by NoVABOT allows users to explore the diagrams and solve related problems; on the other hand, teachers appreciated the capabilities that such a system could provide to students.
In the future, we plan to collect a corpus of interactions similar to [3] to fine-tune an LLM to overcome the limitations recognized in NoVABOT understanding capacity.
Moreover, we plan to experiment the dialogic platform of NovaBOT in order to explain to the users the graphical constraints of formalizing the diet as a Simple Temporal Problem as in [25] in which the authors explored the use of NLG techniques to educate users about maintaining a healthy diet.
During the data collection process, we made sure that all participants in the experiments were thoroughly informed about the research objectives, as well as their rights and responsibilities as participants. Along with the invitation letter, participants received a consent form in which they confirmed, among other things: i) to be aware of the objectives of this research; ii) to participate on a voluntary basis; iii) to be of legal age; iv) to be aware that the study was in line with current data processing and protection regulations, on both national and EU level; v) to be aware of the possibility of withdrawing from the study at any time, without explanation, without any penalty and obtaining the non-use of the data.
This research has been developed for the project NoVAGraphS (Non-Visual Access to Graphical Structure), and is partially supported by the grant Progetto CRT 2021.1930 from Fondazione CRT (Cassa di Risparmio di Torino). [4] M. McTear, Z. Callejas, D. Griol, The Conversational Interface: Talking to Smart Devices, Springer,
Cham, 2016. doi:10.1007/978- 3- 319- 32967- 3. [5] C. Bernareggi, C. Comaschi, G. Dalto, P. Mussio, L. Parasiliti Provenza, Multimodal exploration and manipulation of graph structures, in: Proceedings of the 11th International Conference on Computers Helping People with Special Needs, ICCHP ’08, Springer-Verlag, Berlin, Heidelberg, 2008, p. 934–937. doi:10.1007/978- 3- 540- 70540- 6_140. [6] C. Bernareggi, D. Ahmetovic, S. Mascetti, muGraph: Haptic Exploration and Editing of 3D Chemical Diagrams, in: Proceedings of the 21st International ACM SIGACCESS Conference on Computers and Accessibility, ASSETS ’19, Association for Computing Machinery, New York, NY, USA, 2019, p. 312–317. doi:10.1145/3308561.3353811. [7] A. Frances, V. Scoy, D. Mclaughliny, J. Odomy, R. Wallsz, M. Zuppuhaurô, Touching mathematics: A prototype tool for teaching pre-calculus to visually impaired students, Journal of Modern Optics - J MOD OPTIC 53 (2006). doi:10.1080/09500340600618652. [8] D. Ahmetovic, C. Bernareggi, J. a. Guerreiro, S. Mascetti, A. Capietto, Audiofunctions.web: Multimodal exploration of mathematical function graphs, in: Proceedings of the 16th International Web for All Conference, W4A ’19, Association for Computing Machinery, New York, NY, USA, 2019, pp. 1–10. doi:10.1145/3315002.3317560. [9] S. Mascetti, A. Gerino, C. Bernareggi, L. Picinali, On the evaluation of novel sonification techniques for non-visual shape exploration, ACM Transactions on Accessible Computing 9 (2017) 1–28. doi:10.1145/3046789. [10] J. Su, A. Rosenzweig, A. Goel, E. de Lara, K. N. Truong, Timbremap: enabling the visually-impaired to use maps on touch-enabled devices, in: M. de Sá, L. Carriço, N. Correia (Eds.), Proceedings of the 12th Conference on Human-Computer Interaction with Mobile Devices and Services, Mobile HCI 2010, Lisbon, Portugal, September 7-10, 2010, ACM International Conference Proceeding Series, ACM, 2010, pp. 17–26. doi:10.1145/1851600.1851606. [11] V. Sorge, M. Lee, S. Wilkinson, End-to-end solution for accessible chemical diagrams, in: Proceedings of the 12th International Web for All Conference, W4A ’15, Association for Computing Machinery, New York, NY, USA, 2015. doi:10.1145/2745555.2746667. [12] B. G. Zapirain, A. M. Zorrilla, I. Ruiz, A. Muro, Learning electronics using image processing techniques for describing circuits to blind students, in: The 10th IEEE International Symposium on Signal Processing and Information Technology, 2010, pp. 156–160. doi:10.1109/ISSPIT.2010. 5711764. [13] S. Chockthanyawat, E. Chuangsuwanich, A. Suchato, P. Punyabukkana, Towards automatic diagram description for the blind, in: i-CREATe. The International Convention on Rehabilitation Engineering and Assistive Technology, 2017, pp. 1–4. doi:10.13140/RG.2.2.11969.04961. [14] I. Škrjanec, M. S. Edhi, V. Demberg, Barch: an English dataset of bar chart summaries, in: N. Calzolari, F. Béchet, P. Blache, K. Choukri, C. Cieri, T. Declerck, S. Goggi, H. Isahara, B. Maegaard, J. Mariani, H. Mazo, J. Odijk, S. Piperidis (Eds.), Proceedings of the Thirteenth Language Resources and Evaluation Conference, European Language Resources Association, Marseille, France, 2022, pp. 3552–3560. URL: https://aclanthology.org/2022.lrec-1.380. [15] A. Mazzei, M. Monticone, C. Bernareggi, Using nlg for speech synthesis of mathematical sentences, in: INLG 2019 - 12th International Conference on Natural Language Generation, Proceedings of the Conference, Association for Computational Linguistics (ACL), Stroudsburg, PA, 2019, pp. 463–472. [16] A. Mazzei, M. Monticone, C. Bernareggi, Evaluating speech synthesis on mathematical sentences, in: CEUR Workshop Proceedings, volume 2481, CEUR-WS, BERLIN, 2019, pp. 1–6. URL: https: //ceur-ws.org/Vol-2481/paper46.pdf. [17] P. F. Balestrucci, L. Anselma, C. Bernareggi, A. Mazzei, Building a spoken dialogue system for supporting blind people in accessing mathematical expressions, in: Proceedings of the 9th Italian Conference on Computational Linguistics (CLiC-it 2023), volume 3596, CEUR-WS, Aachen, 2023, pp. 1–8. URL: https://ceur-ws.org/Vol-3596/paper8.pdf. [18] P. F. Balestrucci, E. Di Nuovo, M. Sanguinetti, L. Anselma, C. Bernareggi, A. Mazzei, An educational dialogue system for visually impaired people, IEEE Access 12 (2024) 150502–150519. doi:10.1109/ ACCESS.2024.3479883. [19] H. Bunt, V. Petukhova, D. Traum, J. Alexandersson, Dialogue Act Annotation with the ISO 24617-2 Standard, Springer International Publishing, Cham, 2017, pp. 109–135. doi:10.1007/ 978-3-319-42816-1_6. [20] H. Bunt, V. Petukhova, E. Gilmartin, C. Pelachaud, A. Fang, S. Keizer, L. Prévot, The ISO standard for dialogue act annotation, second edition, in: N. Calzolari, F. Béchet, P. Blache, K. Choukri, C. Cieri, T. Declerck, S. Goggi, H. Isahara, B. Maegaard, J. Mariani, H. Mazo, A. Moreno, J. Odijk, S. Piperidis (Eds.), Proceedings of the Twelfth Language Resources and Evaluation Conference, European Language Resources Association, Marseille, France, 2020, pp. 549–558. URL: https: //aclanthology.org/2020.lrec-1.69. [21] E. Reiter, R. Dale, Building natural language generation systems, 2000. doi:10.1017/ cbo9780511519857. [22] R. Martins, F. Viseu, H. Rocha, Functional thinking: A study with 10th-grade students., Education
Sciences 13 (2023). doi:10.3390/educsci13040335. [23] C. S. Peirce, Collected Papers, The Belknap Press of Harvard Univ. Press, Cambridge, 1965. [24] M. Schrepp, User Experience Questionnaire Handbook. All you need to know to apply the UEQ successfully in your project, https://www.ueq-online.org/, 2015. [25] A. Mazzei, L. Anselma, F. D. Michieli, A. Bolioli, M. Casu, J. Gerbrandy, I. Lunardi, Mobile computing and artificial intelligence for diet management, in: V. Murino, E. Puppo, D. Sona, M. Cristani, C. Sansone (Eds.), New Trends in Image Analysis and Processing - ICIAP 2015 Workshops - ICIAP 2015 International Workshops: BioFor, CTMR, RHEUMA, ISCA, MADiMa, SBMI, and QoEM, Genoa, Italy, September 7-8, 2015, Proceedings, volume 9281 of Lecture Notes in Computer Science, Springer, 2015, pp. 342–349. URL: https://doi.org/10.1007/978-3-319-23222-5_42. doi:10.1007/978-3-319-23222-5\_42.
Subject: Invitation to Participate - Testing a Dialogue System for Accessibility of Finite State Automata, Circuits, Functions, and Sets
We are a research group from the Department of Computer Science and the Polin Lab at the University of Turin.
We need your help to test a dialogue system for exploring finite automata, circuits, sets, and functions. If you decide to assist us, you will interact with a dialogue system that we developed with the aim of improving the accessibility of these topics for visually impaired individuals.
Presentation of the NoVAGraphS Research Experiment Our research team needs the valuable assistance of people with minimal knowledge of at least one of the following topics: finite state automata, circuits, sets, and functions. We need your feedback to evaluate our dialogue system designed to improve the accessibility of these topics, which are typically taught in courses on Computer Science, Physics, and Mathematics.
The crucial point is that finite state automata, sets and functions, and circuits are generally represented by images that are not accessible to visually impaired individuals. For this reason, we have developed NoVABot, a software that allows visually impaired students to engage in dialogue to understand these topics in an equivalent way.
What to Expect? Testing the dialogue system: interact with the prototype of the dialogue system using a web interface to explore an automaton, circuit, sets, and functions.
As experts in at least one of these disciplines, explore the software to understand the benefits students may gain from using a customized system.
Complete a questionnaire on Google Forms: we will ask you a few questions to assess the usability and efectiveness of NoVABot.
Interested in Participating or Need More Information to Decide? To participate or request more information, please contact us at the following email address: pierfelice.balestrucci@unito.it
NoVAGraphS Research Group