How I did customize "rasa-nlu-trainer" as my own tool

Check out my implementation here

Background

I wanted to have a tool for human beings to classify intents and extract entities of texts which were obtained from a raw dataset such as Rocket.chat's conversation, Maluuba Frames or here. Then, the output (labeled texts) could be consumed by an NLU tool such as Rasa NLU.

rasa-nlu-trainer was a potential one which I didn't need to build an app from scratch. However, I needed to add more of my own features to fulfill my needs. They were:

1. Loading/displaying raw texts stored by a database such as MongoDB
2. Manually labeling intents and entities for the loaded texts
3. Persisting labeled texts into the database

I firstly did look up what rasa-nlu-trainer's technologies were used in order to see how to implement my mentioned features.

At first glance

rasa-nlu-trainer was bootstrapped with Create React App. Create React App is a tool to create a React app with no build configuration, as it said. This tool is also recommended by the official React.js tutorial. I actually was very new to React.js. I wondered what problem Creat React App tried to solve so that I first wanted to set up a project without Create React App. I followed this post.

Well, let me say if I built a React app from scratch then I needed to do some manual configurations for Webpack and Babel including webpack.config.json and .babelrc

Then, I tried to use Create React App, and I followed this guide. All right, let me say Create React App helped me create a React boilerplate project with hiding Webpack and Babel's configuration so that as a developer I only needed to focus on writing production code.

The nuts and bolts

I went on starting to watch how rasa-nlu-trainer's source code was organized and what its techniques are used.

Folder structure

1. Auto-created stuff by Create React App
- public/index.html
- src/index.js
2. Overrides
- They didn't use "react-script", they created their own way instead! That was the reason why some additional folder/files were added such as server.js, scripts and config.

Used technologies

- Programming language: Javascript
- Babel: ES6 transpiler
- Project bundle: Webpack
- Dependencies management: npm
- Backend: Node.js
- Fronted framework/project-based: React.js
- React model storing: React-Redux
- React UI extension components: Ant Design
- TypeScript

Hooray! Welcome me to Javascript world!

Architecture

- index.js: Provider{store} <---> LocaleProvider ---> App
- How "separation of concerns" works: import and export
- Global data (store) is immutable and only changed by files actions.js and reducer.js

Proposed changes

Finally, I could start to add my needs to the existing rasa-nlu-trainer app. I listed out what my features:
- Creating an overview page as a custom React component to display all loaded texts
- Customizing the existing page for classifying intents and extracting entities.
- Implementing Node.js methods to load/stores texts (as JSON) which stored in MongoDB

The rule of thumb was I needed to follow the mentioned architecture for modifying the global data which were managed by React-Redux.

You can take a look at the changes from this pull request.

Conclusion

Sometimes, the best way to learn new techs is to get down into an existing tool and customize it with your own needs. 💪

Comments

Attribute 'for' of label component with id xxxx is not defined

I got the warning in the log file when I have used the tag <h:outputLabel> without attribute " for " in xhtml file. It was really polluting my server log files. The logged information actually makes sense anyway! We could find an answer as the following: "Having h:outputLabel without a "for" attribute is meaningless. If you are not attaching the label, you should be using h:outputText instead of h:outputLabel." However, these solutions are not possible just for my situation. Instead of using h:outputText for only displaying text, my team has used h:outputLabel too many places. We were nearly in our release time (next day) so it is quite risky and takes much efforts if we try to correct it. Because the style (with CSS) is already done with h:ouputLabel . The alternative by adding attribute " for " the existing h:outputLabel is not reasonable either. I really need to find another solution. Fortunately, I came across a way if I cha...

What the heck is Meteor DDP?

I was using Meteor for my messenger project. I was so curious about the real time connection. I wanted to know how exactly this mechanism works. In this post, I will go through the DDP Specification, an overview of WebSocket, and a simple demo about how to subscribe a publication of Rocket.Chat (containing a DDP server) from an external webpage. At a glance, I knew that Meteor invented a protocol called DDP which uses for handling real time connection. So then, what is DDP? "DDP (Distributed Data Protocol) is the stateful WebSocket protocol that Meteor uses to communicate between the client and the server." [1] All right! Why does DDP matter? "DDP is a standard way to solve the biggest problem facing client-side JavaScript developers: querying a server-side database, sending the results down to the client, and then pushing changes to the client whenever anything changes in the database" . [2] In order to understand deeply the protocol, I decided ...

JSF 2 - Dynamically manipulating the component tree with system events

Let's suppose we want to modify the metadata (attributes) of elements such as render , requried , maxlength but we do not define in JSF tags. The manipulating components can be conducted in Drools files, for example. How could we do? I think that is what we need to change something of component tree during JSF life-cycle. JSF supports event handling throughout the JSF life-cycle. In this post, I use two events: postAddToView for scanning components tree and preRenderView for manipulating the meta of components before rendering to GUI. I modified my own project from previous post for this example. This is my first further JSF trying out with the project as I said before. :) We define the tags f:event below the form - a container component of the components which we want to work on. The valid values for the attribute type for f:event can be found from tag library document of JSF 2. <!DOCTYPE html> <html xmlns="http://www.w3.org/1999/xhtml" x...

Building a Wizard with Chain of Responsibility Pattern

What is the Idea? We want to create a page that there are some steps and each step has its own business. Users are able to click on a step and its status could be changed. Primefaces owns a component " Wizard " but it it quite hard for us in order to apply our very specific and complicated business domain logic on each step; even we cannot click on a step of this component. We somehow are able to use the component " TabView " works with a strong back-end mechanism. A backend mechanism! what do I mean? Yes, we need it because we want to abstract the behaviors of each step otherwise we will get trouble with many events handling. Obviously, each step has some behaviors such as "next", "back" and "switch' are the same and they are related to each other; but the business of these behaviors can be different totally. That is where the pattern "Chain of Responsibility" can be applied. Step by Step Building It! In this simple pr...

Google I/O '18

Here are my top 5 impressions on this conference. Gmail - live sentence suggestion It looked like the way developers use intelligent code completion in an IDE when coding. Google Photos: converting a photo has a document to pdf I have a paid app on my iPad called "Scanner for Me" but now I can use Google Photos instead. ;) Google Assistant: bot makes a real call to book a restaurant dinner or a hair salon Wow! This feature, for me, is really a big innovation. My team is working so hard on building our bot which is able to have a continued conversation. Google is so good! Google Maps: using Phone's camera to watch the direction When I saw a fox as a cicerone on the demo, I was thinking of Pokémon GO. Google Lens: extract text from images I have heard a story from a friend of me that he had to use his app about "optical character recognition (OCR)" to scan and translate the texts into English whenever he saw texts in China. Google Lens would be ...

vhandit

Search This Blog