---
title: Understanding train section
sidebar_label : Understanding train data
---

The train section provides 7 different NLU training concepts catering to different use cases.

:::note
NLU (a subset of NLP-natural language processing) analyses the text using semantics and understands the meaning of sentences (conversation between a bot and the user).

The same text can have different meanings(and vice versa) or the meaning of a sentence can change depending on the context.

NLU processes input text from the user using computational methods to reach some understanding of the input. For this understanding to work precisely it is necessary for the bot to have a background on what to expect from the conversation. To teach this to the bot is called **training** and is done through the **Train** section.
:::

You can use any or all of the concepts to train your bot. They are namely:
1. [Intents](https://docs.yellow.ai/docs/platform_concepts/studio/train/intents)
2. [Entities](https://docs.yellow.ai/docs/platform_concepts/studio/train/entities)
3. [FAQs](https://docs.yellow.ai/docs/platform_concepts/studio/train/add-faqs)
4. [Documents](https://docs.yellow.ai/docs/platform_concepts/studio/train/what-is-document-cognition)
5. [Synonyms](https://docs.yellow.ai/docs/platform_concepts/studio/train/synonyms)
6. [Small talk ](https://docs.yellow.ai/docs/platform_concepts/studio/train/smalltalk)
7. [Context management](https://docs.yellow.ai/docs/platform_concepts/studio/train/add-contextual-response)


## 1. Training intents

| Pros | Cons |
| -------------------- | ---- |
| Used to start a flow | Cannot be used to answer a specific question |
| Can handle many such similar situations without much training |Will start a whole new flow if trained incorrectly |

## 2. Training entities

| Pros | Cons |
| -------------------- | ---- |
| Used to find **keywords** inside an intent to improve the quality of the response | Harder to train |
| One-time setup for multiple intents |Need to build flows for each entity variant |

## 3. Training FAQs

| Pros | Cons |
| -------------------- | ---- |
| Direct Q&A type of behaviour | Not a smarter option as it requires a lot of training to match real user response |
| Easy to setup | |

## 4. Training Documents

Used for 2 particular use cases. To fetch a response when the bot is reached its **fallback** and to enable search in the flow using the uploaded documents.

## 5. Training Synonyms

You can train multiple synonyms, mainly if they are to the words concerning your use cases.
For example,
> Skin cream: Skin lotion, balm, lotion, serum etc.

## 6. Training Small talk

Few questions from the users like "Hey bot, what is your name?" will try to fetch the response from suggestions/ fallback as per the flow.
If you want the bot users to have a humanly interaction, you can train small talk. With this, casual conversational sentences will not go to the fallbcak (in the bot flow).

Most of the common questions are handled by default.

## 7. Training Contexts