AI Model FAQs
No tags available
Updated
2 years ago
, by [Redesign] Sprinklr Help Center
The industry-specific NLP-based models which are trained on specific industry data to ensure high accuracy are known as AI models. These AI models can also be custom-built for LI clients on their set of data.
The following AI Models are available for Location Insights –
Industry | Model Name |
Automotive | Automobile |
Dentistry | Dental |
Generic | Generic |
Hospitality | Hotel |
Insurance | Insurance |
Food And Beverage | Restaurant |
Retail Bank | Retail Bank |
Retail Store | Retail Store |
Telecommunications | Telecom |
Healthcare (Hospitals) | Healthcare |
The model available in a specific client environment is governed by the contract. To get access to an additional LI model, please get in touch with your Success Manager.
The industry-specific AI Models in Location Insights are trained in multiple languages. Please refer to the KB article for more information on this – Language Support for LI Models.
The AI models are trained on raw/unclassified data which is relevant for the customer with multiple iterations of training which ensures good accuracy for sentiment and AI-driven category classification. Please refer to the KB article for more information on this – How are LI Models Trained?
Different AI models have different out-of-the-box taxonomies which have subject and attribute categories on which the machine is trained for industry specific data. The models have taxonomies which can been referred to in this KB article - Out-of-the-box LI Models
Sprinklr allows the user to add/remove any subject/attribute categories which may or may not be relevant to their use-case. The addition/removal of such categories are driven from the backend. For any such customization request, please contact the Success Team.
Please contact the Success Team for more information on this.
Yes, more languages can be added to any existing Location Insights AI model as per the client requirement. Please contact the Success Team for pricing and feasibility.
Sprinklr ensures an accuracy of more than 80% on its Out-of-the-box AI models. We have a quarterly cycle of re-training the model with new data so as to maintain the accuracy in line with evolving conversation trends.
No, the AI models are trained supervised machine learning methodology. Insights categories which are predicted by the model are predetermined and not keyword based, thus ensuring correct sentiment on phrases within a message.
Yes, it is possible to make a new AI model for an industry on client relevant data from scratch. This will however depend upon the SOW/contract. Please contact the success manager for more details.