.Dialogue system based on spoken conversations with access to an unstructured knowledge base
Annotation
This paper describes an approach for constructing a task-oriented dialog system (a conversational agent) with an unstructured knowledge access based on spoken conversations including: written speech augmentation that simulates the speech recognition results; combination of classifiers; retrieval augmented text generation. The proposed approach provides the training data augmentation in two ways: by converting the original texts into sound waves by a text-to-speech model and then transforming back into texts by an automated speech recognition model; injecting artificially generated errors based on phonetic similarity. A dialogue system with access to the unstructured knowledge base solves the task of detecting a turn, which requires searching for additional information in an unstructured knowledge base. For this purpose, the Support Vector Machine, Convolutional Neural Network, Bidirectional Encoder Representations from Transformers, and Generative Pre-trained Transformer 2 models were trained. The best of the presented models are used in the weighted combination. Next, a suitable text fragment is selected from the knowledge base and a reasonable answer is generated. The tasks are solved by adapting the retrieval augmented text generation model Retrieval Augmented Generation. The proposed method tested on the data from the 10th Dialogue System Technology Challenge. In all metrics, except Precision, the new approach significantly outperformed the results of the basic models proposed by the organizers of the competition. The results of the work can be used to create chat-bot systems that provide automatic processing of user requests in natural language based on an unstructured knowledge access, such as a database of answers to frequently asked questions.
Keywords
Постоянный URL
Articles in current issue
- Characterization of the holographic photopolymer Bayfol HX in the IR region
- Study of blood vessels reaction to local heating by imaging photoplethysmography
- Transmission of 3D holographic information over a radio channel by a method close to SSB
- Anodization parameters influence on anodic aluminum oxide formed above the silver island film
- State estimation accuracy analysis of an induction electric drive by the algorithms of Luenberger and Kalman
- A method of optimizing the structure of hierarchical distributed control systems
- Method for identification of sinusoidal signal parameters with variable unknown amplitude
- Improvement of the automatic temperature stabilisation process in the cryovacuum unit
- Investigation on impact and wear behavior of Al6061 (SiC+Al2O3) and Al7075 (SiC+Al2O3) hybrid composites
- Computational methods to increase the speed of FPGA-based discrete wavelet transforms
- Multiobjective evolutionary discovery of equation-based analytical models for dynamical systems
- Probabilistic criteria for time-series predictability estimation
- Software framework for hyperparameters optimization of models with additive regularization
- Value-based modeling of economic decision making in conditions of unsteady environment
- Methodology for organizing and conducting a study to assess consumer ability
- Automated cluster analysis of communication strategies of educational telegram channels
- Computer modeling of non-Markovianprocesses based on the principle of balance of “complex probabilities”
- The objectification method of the weight coefficients for decision-making in multicriteria problems
- Grishentsev A.Yu., Goroshkov V.A., Chernov R.I.Assessment of the limits of applicability and methods of modulation of near-field magnetic coupling
- Numerical and analytical modeling of the propulsive wing and fuselage of an air taxi
- Using variable-precision feedback to improve operational speed of the current loop in GaN-inverters
- Simulation of diffusion processes during electrothermal treatment of reaction crucibles of the Fe-Sn system