Publications

2025

1. NLP

   Contextual ASR Error Handling with LLMs Augmentation for Goal-Oriented Conversational AI

   Yuya Asano, Sabit Hassan, Paras Sharma, Anthony B. Sicilia, Katherine Atwell, Diane Litman, and Malihe Alikhani

   In Proceedings of the 31st International Conference on Computational Linguistics: Industry Track, Jan 2025

   Abs Bib

   General-purpose automatic speech recognition (ASR) systems do not always perform well in goal-oriented dialogue. Existing ASR correction methods rely on prior user data or named entities. We extend correction to tasks that have no prior user data and exhibit linguistic flexibility such as lexical and syntactic variations. We propose a novel context augmentation with a large language model and a ranking strategy that incorporates contextual information from the dialogue states of a goal-oriented conversational AI and its tasks. Our method ranks (1) n-best ASR hypotheses by their lexical and semantic similarity with context and (2) context by phonetic correspondence with ASR hypotheses. Evaluated in home improvement and cooking domains with real-world users, our method improves recall and F1 of correction by 34% and 16%, respectively, while maintaining precision and false positive rate. Users rated .8-1 point (out of 5) higher when our correction method worked properly, with no decrease due to false positives.

   @inproceedings{asano-etal-2025-contextual,
     title = {Contextual {ASR} Error Handling with {LLM}s Augmentation for Goal-Oriented Conversational {AI}},
     author = {Asano, Yuya and Hassan, Sabit and Sharma, Paras and Sicilia, Anthony B. and Atwell, Katherine and Litman, Diane and Alikhani, Malihe},
     editor = {Rambow, Owen and Wanner, Leo and Apidianaki, Marianna and Al-Khalifa, Hend and Eugenio, Barbara Di and Schockaert, Steven and Darwish, Kareem and Agarwal, Apoorv},
     booktitle = {Proceedings of the 31st International Conference on Computational Linguistics: Industry Track},
     month = jan,
     year = {2025},
     address = {Abu Dhabi, UAE},
     publisher = {Association for Computational Linguistics},
     url = {https://aclanthology.org/2025.coling-industry.32/},
     pages = {374--386},
   }

2. AIED

   Beyond Static Measures: Temporal Analysis of Lexical Alignment in Human-Human Learning With a Teachable Robot

   Paras Sharma, Daniel Fritsch, Yuya Asano, Quentin King-Shepard, Tyree Langley, Tristan Maidment, Diane Litman, Timothy Nokes-Malach, Adriana Kovashka, Nikki Lobczowski, and Erin Walker

   In Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science, Jan 2025

   Abs Bib

   Lexical alignment occurs when conversational partners converge on similar linguistic patterns. In collaborative learning settings, lexical alignment could indicate rapport, which can further predict learning and the collaborators’ evolving shared understanding. Traditional approaches to alignment computation often focus more on the summary statistics computed at the end of the conversation, which usually do not capture the conversational dynamics effectively. This work investigates how alignment evolves in a conversation by modeling lexical alignment trajectories between human dyads while interacting with a teachable robot. We find that, along with the summary statistics, the alignment curve parameters and the time taken to reach key alignment moments significantly predict rapport. We further see significant relationships between the early turns in the conversation and the overall alignment trajectories, indicating the importance of modeling conversational dynamics to plan real-time interventions in the robot, with the goal of altering the alignment trajectories and, consequently, learning outcomes.

   @inproceedings{sharma-etal-2025-temporal-lexical-alignment,
     title = {Beyond Static Measures: Temporal Analysis of Lexical Alignment in Human-Human Learning With a Teachable Robot},
     author = {Sharma, Paras and Fritsch, Daniel and Asano, Yuya and King-Shepard, Quentin and Langley, Tyree and Maidment, Tristan and Litman, Diane and Nokes-Malach, Timothy and Kovashka, Adriana and Lobczowski, Nikki and Walker, Erin},
     editor = {Cristea, A.I. and Walker, E. and Lu, Y. and Santos, O.C. and Isotani, S.},
     booktitle = {Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science},
     volume = {15880},
     year = {2025},
     publisher = {Springer},
     address = {Cham},
     doi = {10.1007/978-3-031-98459-4_20},
     url = {https://doi.org/10.1007/978-3-031-98459-4_20},
   }

3. AIED

   Who’s Got the Power? Data Feminism as a Lens for Designing AIED Engagement Systems

   Angela E.B. Stewart, Jaemarie Solyst, Xinyi Bao, Paras Sharma, Amanda Buddemeyer, Tara Nkrumah, Amy Ogan, and Erin Walker

   In Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science, Jan 2025

   Abs Bib

   A goal of the AIED community is to create equitable systems; yet, we lack a cohesive viewpoint on how to do so. In the present work, we propose power as this organizing principle. We utilize the data feminism framework to showcase how we might balance power, focusing on learner engagement. We utilize multimodal data from ten middle school girls in a virtual computer science camp to discuss how the AIED community might create systems of equity that support all learners.

   @inproceedings{stewart-etal-2025-data-feminism-aied,
     title = {Who's Got the Power? Data Feminism as a Lens for Designing AIED Engagement Systems},
     author = {Stewart, Angela E.B. and Solyst, Jaemarie and Bao, Xinyi and Sharma, Paras and Buddemeyer, Amanda and Nkrumah, Tara and Ogan, Amy and Walker, Erin},
     editor = {Cristea, A.I. and Walker, E. and Lu, Y. and Santos, O.C. and Isotani, S.},
     booktitle = {Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science},
     volume = {15880},
     year = {2025},
     publisher = {Springer},
     address = {Cham},
     doi = {10.1007/978-3-031-98459-4_17},
     url = {https://doi.org/10.1007/978-3-031-98459-4_17},
   }

4. AIED

   Multi-party Lexical Alignment in Collaborative Learning with a Teachable Robot

   Yuya Asano, Diane Litman, Paras Sharma, Daniel Fritsch, Quentin King-Shepard, Timothy Nokes-Malach, Adriana Kovashka, and Erin Walker

   In Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science, Jan 2025

   Abs Bib

   Building rapport with a teachable agent enhances learning. In human-human interactions, speakers build rapport by aligning their conversational behaviors with others. However, the roles of lexical alignment (LA) in building rapport with computational agents are more complex. Computing LA is problematic for emerging multi-party scenarios because neither existing multi-party measures nor combinations of pair-wise measures are designed to model these roles. Thus, we extend an existing LA measure to better capture the dynamics of alignment in multi-party human-computer interactions by automatically extracting lexical patterns used by all speakers and characterizing the alignment behaviors of each (group of) speaker(s). Our new measure predicts rapport in a human-human-robot collaborative scenario better than existing ones and captures individual contributions to a group’s alignment.

   @inproceedings{asano-etal-2025-multi-party-lexical-alignment,
     title = {Multi-party Lexical Alignment in Collaborative Learning with a Teachable Robot},
     author = {Asano, Yuya and Litman, Diane and Sharma, Paras and Fritsch, Daniel and King-Shepard, Quentin and Nokes-Malach, Timothy and Kovashka, Adriana and Walker, Erin},
     editor = {Cristea, A.I. and Walker, E. and Lu, Y. and Santos, O.C. and Isotani, S.},
     booktitle = {Artificial Intelligence in Education. AIED 2025. Lecture Notes in Computer Science},
     volume = {15882},
     year = {2025},
     publisher = {Springer},
     address = {Cham},
     doi = {10.1007/978-3-031-98465-5_16},
     url = {https://doi.org/10.1007/978-3-031-98465-5_16},
   }

2024

1. Robotics Ed, TEL

   Multimodal Sensing of Goals and Activities During Interactions with a Co-created Robot

   Paras Sharma, Veronica Bella, Angela E. B. Stewart, and Erin Walker

   In Technology Enhanced Learning for Inclusive and Equitable Quality Education, Jan 2024

   Abs Bib

   Culturally responsive computing (CRC) curricula engage learners in reflections on power and identity as they build technologies. Open-design tasks, with learner-chosen goals and multiple pathways to achieving them, are common in CRC and could be enhanced using adaptive technologies. Current adaptive technologies function best in well-defined learning trajectories. However, it is unclear how to design these technologies to respond to individual learners’ ideas in open-design settings. In this paper, we prototype a learning system that uses multimodal sensing, log data, and reflective dialogues to build explanatory learner models in open-design settings. We deploy our system in a 2-week summer camp with middle school girls and evaluate the system’s effectiveness to understand learner goals and activities. We show the importance of multiple modalities in making inferences about learner goals and activities.

   @inproceedings{10.1007/978-3-031-72312-4_22,
     author = {Sharma, Paras and Bella, Veronica and E. B. Stewart, Angela and Walker, Erin},
     editor = {Ferreira Mello, Rafael and Rummel, Nikol and Jivet, Ioana and Pishtari, Gerti and Ruip{\'e}rez Valiente, Jos{\'e} A.},
     title = {Multimodal Sensing of Goals and Activities During Interactions with a Co-created Robot},
     booktitle = {Technology Enhanced Learning for Inclusive and Equitable Quality Education},
     year = {2024},
     publisher = {Springer Nature Switzerland},
     address = {Cham},
     pages = {163--169},
     isbn = {978-3-031-72312-4},
   }

2. LLM, AIED

   Designing Simulated Students to Emulate Learner Activity Data in an Open-Ended Learning Environment

   Paras Sharma, and Qichang Li

   In Proceedings of the 17th International Conference on Educational Data Mining, Jul 2024

   Abs Bib

   Open-design environments, under open-ended learning environments, provide high agency to learners to define their goals and pathways toward those goals. However, such environments could be difficult to navigate through for some learners due to this openness in goals and activities. Our long-term goal is to build intelligent pedagogical agents to support learner activities within these environments using different dialogue strategies. In this work, we propose to build a Simulated Students system to emulate learner activities in an open-design environment. We hope to use this simulated data in future work to distill knowledge from Large Language Models (LLMs) to build adaptive, and context-based reinforcement learning dialogue models for learner support. We present the early results and proposed directions of our ongoing work and seek advice on how the different strategies that we propose to use in this work could be further used to build adaptive and context-based dialogue models for effective learning in open-design environments.

   @inproceedings{2024.EDM-doctoral-consortium.122,
     address = {Atlanta, Georgia, USA},
     author = {Sharma, Paras and Li, Qichang},
     booktitle = {Proceedings of the 17th International Conference on Educational Data Mining},
     doi = {10.5281/zenodo.12730023},
     editor = {Paaßen, Benjamin and Epp, Carrie Demmans},
     isbn = {978-1-7336736-5-5},
     month = jul,
     pages = {986--989},
     publisher = {International Educational Data Mining Society},
     title = {Designing Simulated Students to Emulate Learner Activity Data in an Open-Ended Learning Environment},
     year = {2024},
   }

3. EDM

   Building Learner Activity Models From Log Data Using Sequence Mapping and Hidden Markov Models

   Paras Sharma, Angela E.B. Stewart, Qichang Li, Krit Ravichander, and Erin Walker

   In Proceedings of the 17th International Conference on Educational Data Mining, Jul 2024

   Abs Bib

   Open-ended learning environments (OELEs) involve high learner agency in defining learning goals and multiple pathways to achieve those goals. These tasks involve learners transitioning through self-regulated learning (SRL) phases by actively setting goals, applying different strategies for those goals, and monitoring performance to update their strategies. However, because of the flexibility, how learners react to impasses and errors has a critical influence on their learning. An intelligent pedagogical agent (IPA) continuously modeling learner activities could help support learners in these environments. However, this continuous comprehension of behaviors and strategies is difficult in OELEs with evolving goals, ill-defined problem structures, and learning sequences. In this paper, we draw from the literature on SRL phases and cognitive states to investigate the utility of two different methods, Sequence Mapping, and Hidden Markov Models, in building learner activity models from log data collected from a summer camp with 14 middle school girls in an open-design environment. We evaluate the effectiveness of these models separately, and combined, in identifying 7 states: Forethought, Engaged Concentration, Acting, Monitoring, Wheel Spinning, Mind Wandering, and Reflect and Repair. Lastly, we recommend dialogue intervention strategies for an IPA to support learning in OELEs.

   @inproceedings{2024.EDM-short-papers.60,
     address = {Atlanta, Georgia, USA},
     author = {Sharma, Paras and Stewart, Angela E.B. and Li, Qichang and Ravichander, Krit and Walker, Erin},
     booktitle = {Proceedings of the 17th International Conference on Educational Data Mining},
     doi = {10.5281/zenodo.12729890},
     editor = {Paaßen, Benjamin and Epp, Carrie Demmans},
     isbn = {978-1-7336736-5-5},
     month = jul,
     pages = {584--593},
     publisher = {International Educational Data Mining Society},
     title = {Building Learner Activity Models From Log Data Using Sequence Mapping and Hidden Markov Models},
     year = {2024},
   }

2023

1. NLP

   ISABEL: An inclusive and collaborative task-oriented dialogue system

   Anthony Sicilia, Yuya Asano, Katherine Atwell, Qi Cheng, Dipunj Gupta, Sabit Hassan, Mert Inan, Jennifer Nwogu, Paras Sharma, and Malihe Alikhani

   In Alexa Prize TaskBot Challenge 2 Proceedings, Jul 2023

   Bib

   @inproceedings{Pittsburgh2023,
     author = {Sicilia, Anthony and Asano, Yuya and Atwell, Katherine and Cheng, Qi and Gupta, Dipunj and Hassan, Sabit and Inan, Mert and Nwogu, Jennifer and Sharma, Paras and Alikhani, Malihe},
     title = {ISABEL: An inclusive and collaborative task-oriented dialogue system},
     year = {2023},
     url = {https://www.amazon.science/alexa-prize/proceedings/isabel-an-inclusive-and-collaborative-task-oriented-dialogue-system},
     booktitle = {Alexa Prize TaskBot Challenge 2 Proceedings},
   }