A next generation Bot development platform that brings the industry's Bot standards to all major devices. Oscova works offline and doesn't require connecting to any online API. The architecture allows developers to create Bots that take advantage of Dialogs, Intents and Entity recognition with least effort. Oscova also comes with dozens of pre-built system entities backed by intelligent Entity Recognizers for Rapid Bot Development.
For ease of referencing Oscova in .NET project, Oscova is deployed in NuGet and is an integral part of the
The Oscova Bot architecture basically consists of the following elements.
|Dialog||A collection of
|Expression||A pattern or example that defines user input.|
|Intent||A void function marked with
|Context||The current context of conversation.|
|Entity||Parsed elements found within the user statement.|
|Prompt||A message that is sent when a specified entity is missing.|
Natural Language Processing
Oscova's in-built Natural Language Processing engine is designed to fastly decode linguistics features in user input and identify characteristics that best describe an existing Expression attached to an Intent. For this Oscova leverages on advanced AI algorithms that give Oscova the unique ability of scoring Intents with high precision. Due to the modular nature of the NLP engine, Oscova can load WordNet Lexical Database and Word Vectors to perform better semantic analysis and word sense disambiguation.
Oscova NLP engine also comes with in-built support for advanced Machine Learning. When machine learning is enabled Oscova trains itself to understand natural language by building intelligent models that extract relevant features from Expressions provided within Dialogs. As training the bot in Machine Learning mode can take several minutes to hours, the default Processing Mode is always set to Rule.
As some development teams may prefer to run their own Bot servers, Oscova comes with in-built support for online API interface via JSON. Responses and Evaluation Results can easily be serialized with one method call.