Mr. Hemanth Manchabale Papachappa
Staff Software Engineer
Walmart India. Pvt. Ltd.
Semantic Intent Weaver
A Context-Aware Search Algorithm for Ambiguous Query Resolution
Paper ID: 83
9th
International Conference on Computational Intelligence in
Data Science
08-10 January 2026
Outline
• Introduction
• Problem statement and Objective
• SIW Five layer Architecture
• Methodology and Results
• Conclusion and Future Work
• References
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 2/20
Motivation: Ambiguous Queries Need Context
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 3/20
Research Gap and Motivation
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 4/20
Problem Statement & Objective
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 5/20
Five-Layer Architecture (SIW)
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 6/20
Layer 1: Intent Deconstruction
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 7/20
The first layer decomposes the raw query q into a multi-dimensional intent
representation, enabling precise ambiguity detection before retrieval begins.
Layer 2: Context Weaver Engine
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 8/20
The first layer decomposes the raw query q into a multi-dimensional intent
representation, enabling precise ambiguity detection before retrieval begins.
Layer 3: Multi-Dimensional Ranking
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 9/20
Layer 4 and 5: Adaptive Presentation and continues learning
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 10/20
Experiments: Datasets, Metrics & Baselines
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 11/20
Results: Overall Effectiveness (nDCG@10)
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 12/20
Ablation Study & Query-Type Breakdown
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 13/20
Computational Analysis
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 14/20
Response to Reviewer
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 15/20
Reviewer: Question & Response
Q: Q1. Limited real-world applicability.
A: In the absence of full user profiles in public benchmarks, personal and social context
were reconstructed from session-level logs and aggregate interaction statistics. While this
is a standard and reasonable approach for offline evaluation, it may not fully capture the
richness, long-term dynamics, and noise present in real-world user behavior. Future work
will focus on validating SIW under more realistic settings, including datasets with
longitudinal user histories, online evaluations, and privacy-preserving personalization
techniques such as federated or on-device learning.
Response to Reviewer
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 16/20
Reviewer: Question & Response
Q: Computational cost and efficiency trade-offs
A: SIW introduces additional computational overhead compared to ColBERT due to its
dynamic context weaving and multi-dimensional scoring. Although it remains faster than a
full BERT-based re-ranking pipeline, latency may be a concern for large-scale or low-
latency applications. In future work, we plan to explore optimization strategies such as
intent-aware pruning of context dimensions, early-exit mechanisms for low-ambiguity
queries, and model compression or distillation to improve efficiency without significantly
degrading effectiveness.
Response to Reviewer
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 17/20
Reviewer: Question & Response
Q: Limited qualitative analysis
A: The current evaluation primarily emphasizes quantitative metrics and ablation
studies. While these results demonstrate consistent improvements, the paper
includes limited qualitative examples illustrating how SIW resolves ambiguity in
practice. Future revisions will incorporate additional case studies showing
ambiguous queries, inferred intent distributions, dynamically weighted contexts,
and ranked outputs to better illustrate the qualitative advantages of SIW over
baseline methods.
Conclusion and Future Work
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 18/20
Acknowledgements
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 19/20
• We thank the organizing committee of IEEE ICCIDS for the opportunity
to present this work.
• We acknowledge the valuable feedback from reviewers that helped
improve this work.
• We thank our collaborators, colleagues, and institutions for their
support.
References
Mr. Hemanth M P Paper ID:83 ICCIDS 2026 20/20
[1] Clark, K., Luong, M.-T., Le, Q.V., Manning, C.D.: Electra: Pre-training ext encoders as discriminators rather than generators. In: International Conference on Learning Representations
(2020)
[2]Yan, F., Zha, H., Li, F.: Deep reinforced query reformulation for information retrieval. In: Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in
Information Retrieval, pp. 109–118 (2020)
[3] Gao, C., Zhang, W., Liu, B.: Temporal-aware attention mechanisms for dynamic ranking in information retrieval. In: Proceedings of the 47th International ACM SIGIR Conference on
Research and Development in Information Retrieval, pp. 880–890 (2024)
[4] Nogueira, R., Cho, K.: Passage re-ranking with BERT. arXiv preprint arXiv:1901.04085 (2019)
[5] Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., Chen, D., Yih, W.-T.: Dense passage retrieval for open-domain question answering. In: Proceedings of the 2020
Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 6769–6781 (2020)
[6] Khattab, O., Zaharia, M.: ColBERT: Efficient and effective passage search via contextualized late interaction over BERT. In: Proceedings of the 43rd International ACM SIGIR Conference
on Research and Development in Information Retrieval, pp. 39–48 (2020)
[7] Tay, Y., Dehghani, M., Bahri, D., Metzler, D.: Transformer memory as a differentiable search index. In: International Conference on Learning Representations (2022)
[8] Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Nogueira, R., P˘ arvulescu, H., Raline, H., Grave, E., Cohen, W., et al.: Retrieval-augmented generation for knowledge-intensive
nlp tasks. In: Advances in Neural Information Processing Systems, vol. 33, pp. 9459–9474 (2020)
[9] Ma, X., Li, Y., Liu, C., Zhou, Y., Sun, J., Zhang, D.: Query rewriting for retrieval-augmented generation. arXiv preprint arXiv:2305.14283 (2023)
[10] Wang, S., Li, B.Z., Khabsa, M., Fang, H., Ma, H.: Linformer: Self- attention with linear complexity. arXiv preprint arXiv:2006.04768 (2020)
[11] Fan, Z., Liu, Z., Cheng, Z., Liu, J.-G., Pardos, Z.A., He, X.: Graph neural networks for session-based recommendation. IEEE Transactions on Knowledge and Data Engineering (2022)
[12] Sun, Y., Wang, S., Li, Y., Feng, S., Chen, X., Zhang, H., Tian, X.,Zhu, D., Tian, H., Wu, H.: ERNIE: Enhanced Representation through Knowledge Integration. arXiv preprint
arXiv:1904.09223 (2019)
[13] Li, C., Bendersky, M., Kim, Y.-G., Kim, S.-Y., Jose, D., Najork, M.:A neural user model for search personalization. In: Proceedings of the Tenth ACM International Conference on Web
Search and Data Mining,
pp. 631–640 (2017)
[14] Voskarides, N., Meij, E., de Rijke, M.: Query resolution for conversational search with recurrent network models. In: Proceedings of the 43rd international ACM SIGIR conference on
research and development in
information retrieval, pp. 1517–1520 (2020)
[15] An, J., Liu, Z., Sordoni, A., Liu, C., Li, B., Glass, J., Nie, J.-Y.:Unifying conversational search and question answering. In: Proceedings of the 44th International ACM SIGIR Conference
on Research and Development in Information Retrieval (SIGIR), pp. 1431–1441. ACM (2021)
[16] Wu, Y., Liu, A., Li, R., Wang, C., Wang, J.: User-centered conversational recommendation: A survey. In: Proceedings of the 30th ACM international conference on information &
knowledge management, pp.4835–4844 (2021)
[17] Wu, S., Sun, F., Zhang, W., Xie, X., Cui, B.,: Graph neural networks in recommender systems: a survey. In: Proceedings of the 29th ACM international conference on information &
knowledge management, pp. 3523–3532 (2020)
[18] He, B., Wang, Z., Chen, J., Jiang, M., Wang, W.: Learning to match structures of entities. In: Proceedings of The Web Conference 2020, pp.2501–2507 (2020)
[19] He, P., Liu, X., Gao, J., Chen, W.: DeBERTa: Decoding-enhanced BERT with Disentangled Attention. In: Proceedings of the International Conference on Learning Representations
(ICLR) (2021)
[20] Bennett, P.N., Sontag, D., Joachims, T., Collins-Thompson, K., Dumais, S., Radlinski, F.: Inferring and using search session context. In: Proceedings of the fifth ACM international
conference on Web search and Data Mining, pp. 123–132 (2012)
[21] McMahan, B., Moore, E., Ramage, D., Hampson, S., y Arcas, B.A.: Communication-efficient learning of deep networks from decentralized data. In: Artificial intelligence and statistics,
pp. 1273–1282. PMLR (2017)
[22] Sun, Y., Wang, S., Li, Y., Feng, S., Chen, X., Zhang, H., Tian, X.,Zhu, D., Tian, H., Wu, H., Wang, H.: ERNIE: Enhanced Representation
Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 21/20
Thank You!
Questions and discussion are welcome.
hemanthmpkumar123@gmail.com

Semantic Intent Weaver: A Context-Aware Search Algorithm for Ambiguous Query Resolution

  • 1.
    Mr. Hemanth ManchabalePapachappa Staff Software Engineer Walmart India. Pvt. Ltd. Semantic Intent Weaver A Context-Aware Search Algorithm for Ambiguous Query Resolution Paper ID: 83 9th International Conference on Computational Intelligence in Data Science 08-10 January 2026
  • 2.
    Outline • Introduction • Problemstatement and Objective • SIW Five layer Architecture • Methodology and Results • Conclusion and Future Work • References Mr. Hemanth M P Paper ID: 83 ICCIDS 2026 2/20
  • 3.
    Motivation: Ambiguous QueriesNeed Context Mr. Hemanth M P Paper ID:83 ICCIDS 2026 3/20
  • 4.
    Research Gap andMotivation Mr. Hemanth M P Paper ID:83 ICCIDS 2026 4/20
  • 5.
    Problem Statement &Objective Mr. Hemanth M P Paper ID:83 ICCIDS 2026 5/20
  • 6.
    Five-Layer Architecture (SIW) Mr.Hemanth M P Paper ID:83 ICCIDS 2026 6/20
  • 7.
    Layer 1: IntentDeconstruction Mr. Hemanth M P Paper ID:83 ICCIDS 2026 7/20 The first layer decomposes the raw query q into a multi-dimensional intent representation, enabling precise ambiguity detection before retrieval begins.
  • 8.
    Layer 2: ContextWeaver Engine Mr. Hemanth M P Paper ID:83 ICCIDS 2026 8/20 The first layer decomposes the raw query q into a multi-dimensional intent representation, enabling precise ambiguity detection before retrieval begins.
  • 9.
    Layer 3: Multi-DimensionalRanking Mr. Hemanth M P Paper ID:83 ICCIDS 2026 9/20
  • 10.
    Layer 4 and5: Adaptive Presentation and continues learning Mr. Hemanth M P Paper ID:83 ICCIDS 2026 10/20
  • 11.
    Experiments: Datasets, Metrics& Baselines Mr. Hemanth M P Paper ID:83 ICCIDS 2026 11/20
  • 12.
    Results: Overall Effectiveness(nDCG@10) Mr. Hemanth M P Paper ID:83 ICCIDS 2026 12/20
  • 13.
    Ablation Study &Query-Type Breakdown Mr. Hemanth M P Paper ID:83 ICCIDS 2026 13/20
  • 14.
    Computational Analysis Mr. HemanthM P Paper ID:83 ICCIDS 2026 14/20
  • 15.
    Response to Reviewer Mr.Hemanth M P Paper ID: 83 ICCIDS 2026 15/20 Reviewer: Question & Response Q: Q1. Limited real-world applicability. A: In the absence of full user profiles in public benchmarks, personal and social context were reconstructed from session-level logs and aggregate interaction statistics. While this is a standard and reasonable approach for offline evaluation, it may not fully capture the richness, long-term dynamics, and noise present in real-world user behavior. Future work will focus on validating SIW under more realistic settings, including datasets with longitudinal user histories, online evaluations, and privacy-preserving personalization techniques such as federated or on-device learning.
  • 16.
    Response to Reviewer Mr.Hemanth M P Paper ID: 83 ICCIDS 2026 16/20 Reviewer: Question & Response Q: Computational cost and efficiency trade-offs A: SIW introduces additional computational overhead compared to ColBERT due to its dynamic context weaving and multi-dimensional scoring. Although it remains faster than a full BERT-based re-ranking pipeline, latency may be a concern for large-scale or low- latency applications. In future work, we plan to explore optimization strategies such as intent-aware pruning of context dimensions, early-exit mechanisms for low-ambiguity queries, and model compression or distillation to improve efficiency without significantly degrading effectiveness.
  • 17.
    Response to Reviewer Mr.Hemanth M P Paper ID: 83 ICCIDS 2026 17/20 Reviewer: Question & Response Q: Limited qualitative analysis A: The current evaluation primarily emphasizes quantitative metrics and ablation studies. While these results demonstrate consistent improvements, the paper includes limited qualitative examples illustrating how SIW resolves ambiguity in practice. Future revisions will incorporate additional case studies showing ambiguous queries, inferred intent distributions, dynamically weighted contexts, and ranked outputs to better illustrate the qualitative advantages of SIW over baseline methods.
  • 18.
    Conclusion and FutureWork Mr. Hemanth M P Paper ID:83 ICCIDS 2026 18/20
  • 19.
    Acknowledgements Mr. Hemanth MP Paper ID: 83 ICCIDS 2026 19/20 • We thank the organizing committee of IEEE ICCIDS for the opportunity to present this work. • We acknowledge the valuable feedback from reviewers that helped improve this work. • We thank our collaborators, colleagues, and institutions for their support.
  • 20.
    References Mr. Hemanth MP Paper ID:83 ICCIDS 2026 20/20 [1] Clark, K., Luong, M.-T., Le, Q.V., Manning, C.D.: Electra: Pre-training ext encoders as discriminators rather than generators. In: International Conference on Learning Representations (2020) [2]Yan, F., Zha, H., Li, F.: Deep reinforced query reformulation for information retrieval. In: Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 109–118 (2020) [3] Gao, C., Zhang, W., Liu, B.: Temporal-aware attention mechanisms for dynamic ranking in information retrieval. In: Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 880–890 (2024) [4] Nogueira, R., Cho, K.: Passage re-ranking with BERT. arXiv preprint arXiv:1901.04085 (2019) [5] Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., Chen, D., Yih, W.-T.: Dense passage retrieval for open-domain question answering. In: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pp. 6769–6781 (2020) [6] Khattab, O., Zaharia, M.: ColBERT: Efficient and effective passage search via contextualized late interaction over BERT. In: Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 39–48 (2020) [7] Tay, Y., Dehghani, M., Bahri, D., Metzler, D.: Transformer memory as a differentiable search index. In: International Conference on Learning Representations (2022) [8] Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Nogueira, R., P˘ arvulescu, H., Raline, H., Grave, E., Cohen, W., et al.: Retrieval-augmented generation for knowledge-intensive nlp tasks. In: Advances in Neural Information Processing Systems, vol. 33, pp. 9459–9474 (2020) [9] Ma, X., Li, Y., Liu, C., Zhou, Y., Sun, J., Zhang, D.: Query rewriting for retrieval-augmented generation. arXiv preprint arXiv:2305.14283 (2023) [10] Wang, S., Li, B.Z., Khabsa, M., Fang, H., Ma, H.: Linformer: Self- attention with linear complexity. arXiv preprint arXiv:2006.04768 (2020) [11] Fan, Z., Liu, Z., Cheng, Z., Liu, J.-G., Pardos, Z.A., He, X.: Graph neural networks for session-based recommendation. IEEE Transactions on Knowledge and Data Engineering (2022) [12] Sun, Y., Wang, S., Li, Y., Feng, S., Chen, X., Zhang, H., Tian, X.,Zhu, D., Tian, H., Wu, H.: ERNIE: Enhanced Representation through Knowledge Integration. arXiv preprint arXiv:1904.09223 (2019) [13] Li, C., Bendersky, M., Kim, Y.-G., Kim, S.-Y., Jose, D., Najork, M.:A neural user model for search personalization. In: Proceedings of the Tenth ACM International Conference on Web Search and Data Mining, pp. 631–640 (2017) [14] Voskarides, N., Meij, E., de Rijke, M.: Query resolution for conversational search with recurrent network models. In: Proceedings of the 43rd international ACM SIGIR conference on research and development in information retrieval, pp. 1517–1520 (2020) [15] An, J., Liu, Z., Sordoni, A., Liu, C., Li, B., Glass, J., Nie, J.-Y.:Unifying conversational search and question answering. In: Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), pp. 1431–1441. ACM (2021) [16] Wu, Y., Liu, A., Li, R., Wang, C., Wang, J.: User-centered conversational recommendation: A survey. In: Proceedings of the 30th ACM international conference on information & knowledge management, pp.4835–4844 (2021) [17] Wu, S., Sun, F., Zhang, W., Xie, X., Cui, B.,: Graph neural networks in recommender systems: a survey. In: Proceedings of the 29th ACM international conference on information & knowledge management, pp. 3523–3532 (2020) [18] He, B., Wang, Z., Chen, J., Jiang, M., Wang, W.: Learning to match structures of entities. In: Proceedings of The Web Conference 2020, pp.2501–2507 (2020) [19] He, P., Liu, X., Gao, J., Chen, W.: DeBERTa: Decoding-enhanced BERT with Disentangled Attention. In: Proceedings of the International Conference on Learning Representations (ICLR) (2021) [20] Bennett, P.N., Sontag, D., Joachims, T., Collins-Thompson, K., Dumais, S., Radlinski, F.: Inferring and using search session context. In: Proceedings of the fifth ACM international conference on Web search and Data Mining, pp. 123–132 (2012) [21] McMahan, B., Moore, E., Ramage, D., Hampson, S., y Arcas, B.A.: Communication-efficient learning of deep networks from decentralized data. In: Artificial intelligence and statistics, pp. 1273–1282. PMLR (2017) [22] Sun, Y., Wang, S., Li, Y., Feng, S., Chen, X., Zhang, H., Tian, X.,Zhu, D., Tian, H., Wu, H., Wang, H.: ERNIE: Enhanced Representation
  • 21.
    Mr. Hemanth MP Paper ID: 83 ICCIDS 2026 21/20 Thank You! Questions and discussion are welcome. hemanthmpkumar123@gmail.com