Natural Language Processing

NLP

Affinsys is the most comprehensive CX & domain NLP engine, powered by a knowledge graphontology with pre-defined products, attributes, slang, abbreviations and language packs.

Bidirectional Context

Product – Attribute - Product mapping for human like context specific conversations with personalized recommendations.

Polyglot Language Model

Understand mixed- language, slang etc. in the same model across languages.

Sentiment Detection

Superior sentiment detection with subjectivity analysis, intensifiers & negation handling.

Multi-intent processing

Context driven Multi-level intent analysis and joint slot filling.

Affinsys Natural Language Processing

Shift from “conversational AI” to “Intelligent conversations”

Affinsys NLP powers our prebuilt Multi-lingual chatbots (voice & text).

-

Informational bots with most comprehensive FAQ’s with wide functional coverage across industries like banking, insurance, telecom, retail and more.

-

Transactional /Task bots for transactions, service requests and complaints.

-

Advisory bots for proactive service, product recommendations, persona based offers, cross sell and upsell.

Bidirectional Context

Our NLP model has embedded domain expertise in the knowledge graph ontology which enables machine reasoning to identify context, which forms the basis for intelligent recommendations when paired with our recommendation engine.

Amigo deep learning model has been trained to predict the user intent and ask queries to the customer to get the part of incomplete information for human-like accuracy.

Polyglot Language Model

Affinsys NLP offers the unique capability to handle multiple languages within the same conversation e.g. a customer is chatting in default language French and changes to English midway, the bot will understand the query and give the response in French.

Our pre-trained multi-lingual domain-specific NLP model has built in synonyms, abbreviations, slang, variants of CX queries, service requests and word combinations of commonly asked questions.

Sentiment Detection

Built-in sentiment detection with specific focus on customer experience journeys.

Pre-built emotion detection at multi-levels based on sentence analysis to derive context and emotion.

Our NLP comes with pre-defined (domain-specific) actions based on the emotion detection as per Plutchik Wheel of Emotions e.g. handoff to human agent.

Multi X processing

Our NLP model is designed from scratch for specific industry to handle multiple intents, common spelling errors, letter case, slang, abbreviations, mobile and desktop keyboard typing errors, speech phonetics and dialects.