Technical Architecture

Smart Answers, Quick Solutions: Tap the Information with Query Insight!

Query Insight (QI) is an innovative platform that answers all kinds of questions quickly and accurately with the help of artificial intelligence by leveraging selected sources. Unlike traditional artificial intelligence systems, this platform uses the Agentic approach that interacts with various tools and resources, rather than aiming to respond directly.

Agentic Approach with Query Insight

The agentic approach is based on designing AI agents that can exhibit human-like intelligence and behavior and working together in harmony. QI AI agents can sense their environment and interact with users or other systems. They can make decisions or take actions in accordance with these interactions. They can develop strategies to achieve their goals. This approach expands the ability of AI agents to handle complex tasks, increasing the use of LLMs (Large Language Models).

Self-Management: AI agents can make decisions and take actions on their own.

Reaction: AI agents can react to external stimuli and changes.

Planning: AI agents break complex tasks into manageable steps. Determines the sequence of actions necessary to achieve a goal and adapts plans when faced with challenges.

Proactivity: AI agents can foresee future events and make plans to achieve their goals.

Control and Learning: AI agents evaluate their own performance, learn from mistakes, and improve their approaches over time.

Socialization: AI agents can interact with other agents or systems.

Chain of Thinking: This technique encourages creative problem-solving ability by allowing AI agents to explore different ways of reasoning.

By using the Agentic structure and the “Retrieval Augmented Generation” (RAG) structure together, Query Insight provides the most accurate answers to questions with AI agents that can exhibit complex behaviors and dynamically adapt to environments. It also combines the strengths of generative, discriminative, and Retrieval-Based models.

Query Insight can perform complex tasks by working within the framework of the Large Language Model (LLM). It works by combining the abilities of multiple AI agents. Each AI agent is specialized in specific task areas, can communicate with each other and share information, improving overall operating efficiency.

Productive Question Answering

Generative Question Answering is a technology used in the fields of natural language processing (NLP) and artificial intelligence (AI), specifically generating natural language answers to questions asked by users. This technology is accomplished using generative models, often powered by methods such as deep learning and neural networks.

This method is based on the idea that a machine can understand natural language and learn to generate natural language text in a way that provides a correct answer in terms of grammar and meaning. Generative Question Answering, unlike traditional fixed-answer systems, can generate new and original answers depending on the question asked. This means that the model can understand the context of the question and generate an appropriate response, rather than choosing from predefined answers in the database.

Generative QA systems often use deep learning techniques that learn from large amounts of text data and can use that information to answer new questions. These systems are supported by techniques used in a variety of natural language processing (NLP) tasks, such as text summarization, text generation, and natural language understanding.

The advantages of Productive Question Answering include the ability for users to obtain information in a more natural and interactive manner, the ability to respond to a broader and more diverse range of question types, and the capacity to produce detailed and descriptive answers about a particular topic.

Generative Model

This model is a technology used in the fields of natural language processing (NLP) and artificial intelligence (AI), especially that produces natural language answers to questions asked by users. This technology is often achieved using generative models powered by methods such as deep learning and neural networks.

Retrieval-Based Model

This model is designed to retrieve the most relevant information from a data set or collection of documents. It uses keyword matching and semantic search techniques to find documents or pieces of information most relevant to the question.

Discriminative Model

This model is trained to choose the most accurate answer from a set of potential answers. It determines the most appropriate answer by evaluating the harmony between the question and the answers. It becomes easier to reach correct answers by using all approaches simultaneously. Query Insight can first retrieve relevant information with a retrieval-based model, then generate detailed answers based on this information with a generative model, and finally select the most appropriate answer with a discriminant model.