Dialog-flow, the easy chatbot creation platform.
I have started my internship as a trainee for AI development, so the task that I have been given today was to implement a simple chatbot using platforms like dialog flow. So what is dialog flow, it’s a framework which allows users to train a chatbot easily and effectively based on their needs. for this purpose, you don’t need to have a deep knowledge of NLP or coding.
first of all what you have to do is to create an account with Dialogflow, which will be much easy as it contains sign up with Google option as well ( remember Google owns this site). After signing up you have to create a dialogflow agent, here the agent will be the model that you will be training as per your need.
On the following page, you will be prompted to select in which language you want to train your model and some other questions.
After finishing creating your model, you have to understand clearly the following three terms,
- Intents
- Entities
- Parameters
for each conversation cycle, you have to create a unique intent, which will run to collect the parameters within the query. So what are parameters? parameters are the important and relevant words which will be used by the model from the users’ query to provide a proper response. (kind of parameters are keywords the model is searching for).
for each intent in the training phase you have to provide what the system will be expecting from the user and how should it respond for that intent. If parameters are found in the input those are also should be marked.
In the given example above the entity language is systems default, so what if we have to create our own entity? You can create a new entity by clicking ‘+’ next to entity. You have to give a name to your entity, in the text field below you can add parameters that belonging to your entity and give any number of synonyms as possible for each entry, then save the system.
While saving your model you can see a gear wheel rotating by the side of the name of your agent, indicating that your model is being trained.
If we take a quick recap of our terms we used before proceeding to next step. Intents are the way by which the needed information is queried from the user. Entities are the knowledge base of the model where the critical data are saved, entities work in collaboration with intent to provide a necessary response to the user.
In the entities we can also restrict the user to enter the necessary details compulsorily. There is also a feature called fullfilment request, which prompts the user to enter the missing compulsory entities with the use of webhooks. We will look into that topic in a separate post.
Basically there are the steps needed to create a chatbot. After that you can integrate the bot with any applications like Google assistant, messenger, vibe etc.
