Since 2011, we were working on creating CAILA – our own Natural Language Understanding service, so that now you could use its powerful capabilities to the max
An end-to-end solution integrated with all Just AI products
Machine learning algorithms ease the process of working with datasets and building named entity recognition and intent classification models
Multilanguage support: English, Russian, Chinese, Portuguese, Japanese
CAILA is the backbone of our products
CAILA is the backbone of our products
Aimylogic
A simple yet feature-packed builder to create AI-powered solutions
Just AI Conversational Platform
A full-range and secure conversational AI platform for solutions of any complexity. Keep everything under control with key statistics and customized reports
Just AI Conversational Framework
An open-source Kotlin-based conversational framework that is free of charge even for commercial use
Why CAILA
CAILA is built into Just AI products. No need to integrate with a third-party NLU service or search for libraries:
- Qualitative text processing: recognition of synonyms, acronyms, and slang expressions
- Increased accuracy of intent recognition, faster processing of unlabeled logs and data selection to train AI systems
- New NLP models and algorithms powered by machine learning, rule-based approach, semantic and morphological analysis
- Provided access to Just AI’s proprietary language technologies and linguistic libraries containing source code
How does CAILA work
Someone is asking a chatbot to make a call
CAILA defines the subject of inquiry and all the characteristics mentioned by the person.
Read documentation
A handy solution for professionals
- Computational linguists
- NLP engineers
- Data scientists
Make the most of CAILA
Intent recognition
- ML and deep learning
- Semantic Textual Similarity algorithm
- Rule-based approach
NLU model debug toolset:
- Visualizing clustering result based on semantic distance
- Model's confidence level analysis and threshold setting
- Word alignment
NLU models enhancement
- Automatic spell-checking based on system dictionaries
- Custom dictionaries for spell checking and misprints
Dialog labeling
- Text preparation (stop words removal, normalization, etc)
- Keywords retrieval
- Phrase clustering by k-means and other ml algorithms
Slots markup and filling:
- Multiple slots in utterance
- Slot filling sub-scenarios
Supported languages
- English, Russian, Chinese, Portuguese, Japanese
- Any other language can be added by request
Third-party NLU algorithms API:
- Agent assistant built on client’s logs
- Integration with dictionaries from external systems
- FAQ knowledge base
- Combination of intent recognition models for better performance
Named entity extraction:
- Duckling entities
- Dictionaries (custom & system)
- Rule-based approach (regexp alike)