| Autor: |
J M Prajwal, M. V. Manoj Kumar, Shyamanth Kashyap, Pavan R. Nargund, G Rahul |
| Rok vydání: |
2019 |
| Předmět: |
|
| Zdroj: |
SSRN Electronic Journal. |
| ISSN: |
1556-5068 |
| DOI: |
10.2139/ssrn.3511387 |
| Popis: |
This paper presents a method for constructing a software robot “CAISY” for answering queries for any training set(domain). This has been achieved with the help of word embedding and Sequence 2 Sequence model using Long Short Term Memory neural networks. The proposed method can respond to test queries under a maximum of 5 seconds response time. The exciting feature of Caisy is it could be applied to any domain where the data is available for training the model. For the demonstration, this paper presents Caisy is taught for answering queries related to Normal Conversation. Caisy has been optimized by minimizing the loss value with the help of variation of hyperparameters. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
