SUSI AI 5 Star Skill Rating System

For making a system more reliable and robust, continuous evaluation is quite important. So is in case of SUSI AI. User feedback is important to improve SUSI skills and create new ones. Previously we had only thumbs up / thumbs down as a feedback method, from the susi chat client. But now a 5 star rating system has been added to the SUSI Skill CMS so that users can rate a skill there. Before the implementation of API  let’s look how data is stored in SUSI AI Susi_server uses DAO in which skill rating is stored as JSONTray.

The server side implementation

A new java class has been created for the API, FiveStarRateSkillService.java.

public class FiveStarRateSkillService extends AbstractAPIHandler implements APIHandler {
    private static final long serialVersionUID =7947060716231250102L;
    @Override
    public BaseUserRole getMinimalBaseUserRole() {
        return BaseUserRole.USER;
    }
    @Override
    public JSONObject getDefaultPermissions(BaseUserRole baseUserRole) {
        return null;
    }
    @Override
    public String getAPIPath() {
        return "/cms/rateSkill.json";
    }
...
}

The getMinimalBaseRole method tells the minimum User role required to access this servlet it can also be ADMIN, USER or ANONYMOUS. In our case it is USER. A user needs to be logged in to rate a skill on a scale of 1-5 stars.  The API runs at “/cms/fiveStarRateSkill.json” endpoint.

Next, create serviceImpl method in the above class to handle the request from the client and respond to it.

  1. Fetch the required query parameters and store them in variables. They include skill model, group, language, skill name and starts that the user has given in the rating.

String skill_name = call.get("skill", null);
String skill_stars = call.get("stars", null);
  1. Then check if the skill exists. If not them throw an exception. Otherwise, increment the count of the corresponding rating. The rating object has keys as one_star, two_star, three_star, four_star and five_star that has the count of that star rating.       

if (skill_stars.equals("1")) {
    skillName.put("one_star", skillName.getInt("one_star") + 1 + "");
}
else if (skill_stars.equals("2")) {
    skillName.put("two_star", skillName.getInt("two_star") + 1 + "");
}
else if (skill_stars.equals("3")) {
    skillName.put("three_star", skillName.getInt("three_star") + 1 + "");
}
else if (skill_stars.equals("4")) {
    skillName.put("four_star", skillName.getInt("four_star") + 1 + "");
}
else if (skill_stars.equals("5")) {
    skillName.put("five_star", skillName.getInt("five_star") + 1 + "");
}
  1. Re-calculate the total rating done on that skill and its average rating and update the object. If the skill has not been already rated then create a new rating object and initialize it with the 0 star counts.

public JSONObject createRatingObject(String skill_stars) {
        JSONObject skillName = new JSONObject();
        JSONObject skillStars = new JSONObject();

        skillStars.put("one_star", 0);
        skillStars.put("two_star", 0);
        skillStars.put("three_star", 0);
        skillStars.put("four_star", 0);
        skillStars.put("five_star", 0);
        skillStars.put("avg_star", 0);
        skillStars.put("total_star", 0);

        skillName.put("stars", skillStars);
}

The complete FiveStarRateSkillService.java is available here : –

https://github.com/fossasia/susi_server/blob/development/src/ai/susi/server/api/cms/FiveStarRateSkillService.java

Rating a skill

Sample endpoint

https://api.susi.ai/cms/fiveStarRateSkill.json?model=general&group=Knowledge&language=en&skill=aboutsusi&stars=3&callback=p&_=1526813916145

This gives 3 star rating to the “aboutsusi” skill.

Parameters

  • Model
  • Group
  • Language
  • Skill
  • Stars

Response

{
  "ratings": {
    "one_star": 0,
    "four_star": 0,
    "five_star": 1,
    "total_star": 1,
    "three_star": 0,
    "avg_star": 5,
    "two_star": 0
  },
  "session": {"identity": {
    "type": "email",
    "name": "[email protected]",
    "anonymous": false
  }},
  "accepted": true,
  "message": "Skill ratings updated"
}

Getting the stats of Skill Ratings

Sample endpoint

http://127.0.0.1:4000/cms/getSkillRating.json?model=general&group=Knowledge&language=en&skill=aboutsusi&callback=p&_=1526813916145

This fetches the current ratings of the “aboutsusi” skill.

Parameters

  • Model
  • Group
  • Language
  • Skill

Response

{
    "session": {
        "identity": {
            "type": "host",
            "name": "172.68.144.159_81c88a10",
            "anonymous": true
        }
    },
    "skill_name": "aboutsusi",
    "accepted": true,
    "message": "Skill ratings fetched",
    "skill_rating": {
        "negative": "0",
        "positive": "0",
        "stars": {
            "one_star": 0,
            "four_star": 2,
            "five_star": 1,
            "total_star": 4,
            "three_star": 1,
            "avg_star": 4,
            "two_star": 0
        },
        "feedback_count": 3
    }
}

Conclusion

So this 5 star rating system will help in improving the SUSI skills. Also, it will help in making better decisions when we have multiple similar skills and we have to choose one to respond to the user query.

References

Post feedback for SUSI Skills in SUSI iOS

SUSI iOS, web and Android clients allow the user to rate the SUSI Skills in a 5-star rating system. Users can write about how much particular skill is helpful for them or if improvements are needed. Users can rate skills from one to five star as well. Here we will see how to submit feedback for SUSI skills and how it is implemented on SUSI iOS.

How to submit feedback –

  1. Go to Skill Listing Screen < Skill Detail Screen
  2. Scroll to the feedback section
  3. Write feedback about SUSI skill
  4. Click on POST button to post the skill feedback

An anonymous user can not submit skill feedback. You must have to logged-in in order to post skill feedback. If you are not logged-in and click POST button to post skill feedback, an alert is presented with Login option, by clicking Login, the user is directed to Login screen where the user can log in and later can post skill feedback.

Implementation of posting skill feedback –

Google’s Material textfield is used for skill feedback text field. We have assigned TextField class from Material target to skill feedback text field to make it very interactive and give better user experience.

Skill feedback text field in the normal state –

Skill feedback text field in the active state –

When the user clicks POST after writing skill feedback, we check if the user is logged-in or not.

if let delegate = UIApplication.shared.delegate as? AppDelegate, let user = delegate.currentUser {
...
}

We have saved the logged-in user globally using AppDelegate shared method during login and using it here. The AppDelegate is sort of like the entry point for the application. It implements UIApplicationDelegate and contains methods that are called when application launches, when is going to the background (i.e. when the user hit the home key), when it’s opened back up, and more. The AppDelegate object is stored as a property on the UIApplication class and is accessible from anywhere in swift classes.

Case 1: If the user is not logged-in, we show a popup to the user with the login option

By clicking Login, the user is directed to Login screen where the user can log in and later can post skill feedback.

Case 2: If the user is already logged-in, we use the endpoint below for posting skill feedback –

http://api.susi.ai/cms/feedbackSkill.json

ModelWith following parameters –

  • Group
  • Skill
  • Feedback
  • Access token
Client.sharedInstance.postSkillFeedback(postFeedbackParam) { (feedback, success, responseMessage) in
DispatchQueue.main.async {
if success {
self.skillFeedbackTextField.text = ""
self.skillFeedbackTextField.resignFirstResponder()
self.view.makeToast(responseMessage)
} else {
self.view.makeToast(responseMessage)
}
}
}

In return response, we get feedback posted by the user –

{
feedback: "Helpful",
session:
{
...
},
accepted: true,
message: "Skill feedback updated"
}

 

Resources –

  1. Material Design Guidelines for iOS
  2. Apple’s documentation on UIApplicationDelegate API
  3. Apple’s documentation on UIApplication API
  4. ChrisRisner’s article on Singletons and AppDelegate

Skill Development using SUSI Skill CMS

There are a lot of personal assistants around like Google Assistant, Apple’s Siri, Windows’ Cortana, Amazon’s Alexa, etc. What is then special about SUSI.AI which makes it stand apart from all the different assistants in the world? SUSI is different as it gives users the ability to create their own skills in a Wiki-like system. You don’t need to be a developer to be able to enhance SUSI. And, SUSI is an Open Source personal assistant which can do a lot of incredible stuff for you, made by you.

So, let’s say you want to create your own Skill and add it to the existing SUSI Skills. So, these are the steps you need to follow regarding the same –

  1. The current SUSI Skill Development Environment is based on an Etherpad. An Etherpad is a web-based collaborative real-time editor. https://dream.susi.ai/ is one such Etherpad. Open https://dream.susi.ai/ and name your dream (in lowercase letters).
  2. Define your skill in the Etherpad. The general skill format is

::name <Skill_name>
::author <author_name>
::author_url <author_url>
::description <description> 
::dynamic_content <Yes/No>
::developer_privacy_policy <link>
::image <image_url>
::term_of_use <link>

#Intent
User query1|query2|query3....
Answer answer1|answer2|answer3...

 

Patterns in query can be learned easily via this tutorial.

  1. Open any SUSI Client and then write dream <your dream name> so that dreaming is enabled for SUSI. Once dreaming is enabled, you can now test any skills which you’ve made in your Etherpad.
  2. Once you’ve tested your skill, write ‘stop dreaming’ to disable dreaming for SUSI.
  3. If the testing was successful and you want your skill to be added to SUSI Skills, send a Pull Request to susi_skill_data repository providing your dream name.

How do you modify an existing skill?

SUSI Skill CMS is a web interface where you can modify the skills you’ve made. All the skills of SUSI are directly in sync with the susi_skill_data.

To edit any skill, you need to follow these steps –

  1. Login to SUSI Skill CMS website using your email and password (or Sign Up to the website if you haven’t already).
  2. Click on the skill which you want to edit and then click on the “edit” icon.
  3. You can edit all aspects of the skill in the next state. Below is a preview:

Make the changes and then click on “SAVE” button to save the skill.

What’s happening Behind The Scenes in the EDIT process?

  • SkillEditor.js is the file which is responsible for keeping a check over various validations in the Skill Editing process. There are certain validations that need to be made in the process. Those are as follows –
  • Check whether User has logged in or not

if (!cookies.get('loggedIn')) {
            notification.open({
                message: 'Not logged In',
                description: 'Please login and then try to create/edit a skill',
                icon: <Icon type='close-circle' style={{ color: '#f44336' }} />,
            });
            this.setState({
                loading: false
            });
            return 0;
        }

 

  • Check whether Commit Message has been entered by User or not

if (this.state.commitMessage === null) {
            notification.open({
                message: 'Please add a commit message',
                icon: <Icon type='close-circle' style={{ color: '#f44336' }} />,
            });

            this.setState({
                loading: false
            });
            return 0;
        }

 

  • Check to ensure that request is sent only if there are some differences in old values and new values

if (this.state.oldGroupValue === this.state.groupValue &&
          this.state.oldExpertValue === this.state.expertValue &&
          this.state.oldLanguageValue === this.state.languageValue &&
          !this.state.codeChanged && !this.state.image_name_changed) {
            notification.open({
                message: 'Please make some changes to save the Skill',
                icon: <Icon type='close-circle' style={{ color: '#f44336' }} />,
            });
            self.setState({
                loading: false
            });
            return 0;
        }

 

  • After doing the above validations, a request is sent to the Server and the User is shown a notification accordingly, whether the Skill has been uploaded to the Server or there has been some error.

$.ajax(settings)
            .done(function (response) {
                this.setState({
                    loading: false
                });
                let data = JSON.parse(response);
                if (data.accepted === true) {
                    notification.open({
                        message: 'Accepted',
                        description: 'Your Skill has been uploaded to the server',
                        //success/>
                    });
                }
                else {
                    this.setState({
                        loading: false
                    });
                    notification.open({
                        message: 'Error Processing your Request',
                        description: String(data.message),
                        //failure />
                    });
                }
            }

 

  • If the User is notified with a Success notification, then to verify whether the Skill has been added or not, the User can go to susi_skill_data repo and see if he has a recent commit regarding the same or not.

Resources

Making a SUSI Skill to get details about bank from IFSC

We are going to make a SUSI skill that fetches information about a bank when the IFSC (Indian Financial System Code) is known. Here is a detailed explanation of how we going about doing this.

Getting started with the skill creation

API endpoint that returns the bank details

Before going to the skill development, we need to find an API that would return the bank details from the IFSC, On browsing through various open source projects. I found an apt endpoint by Razorpay. Razorpay is a payment gateway for India which allows businesses to accept, process and disburse payments with ease. The Github link  to the repository is https://github.com/razorpay/ifsc.

API endpoint –  https://ifsc.razorpay.com/<:ifsc>
Request type –  GET
Response type –  JSON

Now, head over to the SUSI Etherpad, which is the current SUSI Skill Development Environment and create a new Pad. 

Here, we need to define the skill in the Etherpad. We will now write rules/intents for the skill. An intent represents an action that fulfills a user’s spoken request.

Intents consist of 2 parts –

  • User query – It contains different patterns of query that user can ask.
  • Answer – It contains the possible answer to the user query.

The main intent that our skill focuses on is, returning the bank name and address from the IFSC code. Here is how it looks –

Name of bank with IFSC code * | Bank's name with IFSC code *
!example:Name of bank with IFSC code SBIN0007245
!expect: The name of bank is State Bank of India
!console:The name of bank with IFSC code $1$ is $object$
{
"url":"https://ifsc.razorpay.com/$1$",
"path":"$.BANK"
}
eol

Part-wise explanation of the intent

  • The first line contains the query pattern that the user can use while querying. You can see that a wildcard character (*) is used in the pattern. It contains the IFSC of the bank that we wish to know, and will later on use to fetch the details via the API.
  • The second line contains an example query, followed by third line that contains the expected answer.
  • Last part of the rule contains the answer that is fetched from an external API –  https://ifsc.razorpay.com/<:ifsc>  ,via the console service  provided by SUSI Skills. Here, <:ifsc> refers to the IFSC that the user wants to know about. We get it from the user query itself, and can access it by the variable name $1$ as it matches with the 1st wildcard present in the query. If there would be 2 wildcards, we could have accessed them by $1$ and $2$ respectively.
  • The console service provides us with an option to enter the url of the API that we want to hit and path of the key we want to use.

The sample response of the endpoint looks like this :

{
  "BANK": "Karnataka Bank",
  "IFSC": "KARB0000001",
  "BRANCH": "RTGS-HO",
  "ADDRESS": "REGD. & HEAD OFFICE, P.B.NO.599, MAHAVEER CIRCLE, KANKANADY, MANGALORE - 575002",
  "CONTACT": "2228222",
  "CITY": "DAKSHINA KANNADA",
  "RTGS": true,
  "DISTRICT": "MANGALORE",
  "STATE": "KARNATAKA"
}

 

  • Since, we want to extract the name of the bank, the BANK key contains our desired value and we will use $.BANK in the path of the console service. And it can be accessed by $object$ in the answer. We frame the answer using $object$ and $1$ variables, and it like the one mentioned in the expected answer. eol marks the end of the console service.
  • Similarly, the intent that gives us the address of the bank looks like this –
Address of bank with IFSC code * | Bank's address with IFSC code *
!example:Address of bank with IFSC code SBIN0007245
!expect: The address of bank is TILAK ROAD HAKIMPARA, P.O.SILIGURI DARJEELING, WEST BENGAL ,PIN - 734401
!console:The address of bank with IFSC code $1$ is $object$
{
  "url":"https://ifsc.razorpay.com/$1$",
  "path":"$.BANK"
}
eol

Testing the skill

  • Open any SUSI Client and then write dream <your dream name> so that dreaming is enabled for SUSI. We will write down dream ifsc. Once dreaming is enabled, you can now test any skills which you’ve made in your Etherpad.
  • We can test the skills by asking queries and matching it with the expected answer. Once the testing is done, write stop dreaming to disable dreaming for SUSI.

  • After the testing was successful completely, we will go ahead and add it to the susi_skill_data.
  • The general skill format is –
::name <Skill_name>
::author <author_name>
::author_url <author_url>
::description <description> 
::dynamic_content <Yes/No>
::developer_privacy_policy <link>
::image <image_url>
::term_of_use <link>

#Intent
User query1|query2|query3....
Answer answer1|answer2|answer3...

We will add the basic skill details and author details to the etherpad file and make it in the format as mentioned above. The final text file looks like this –

::name IFSC to Bank Details
::author Akshat Garg
::author_url https://github.com/akshatnitd
::description It is a bank lookup skill that takes in IFSC code from the user and provides you all the necessary details for the Bank. It is valid for banks in India only
::dynamic_content Yes
::developer_privacy_policy 
::image images/download.jpeg
::terms_of_use 

Name of bank with IFSC code * | Bank's name with IFSC code *
!example:bank with IFSC code *
!expect: The name of bank is SBI
!console:The name of bank with IFSC code $1$ is $object$
{
"url":"https://ifsc.razorpay.com/$1$",
"path":"$.BANK"
}
eol

Address of bank with IFSC code * | Bank's address with IFSC code *
!example:Address of bank with IFSC code *
!expect: The address of bank is 
!console:The address of bank with IFSC code $1$ is $object$
{
"url":"https://ifsc.razorpay.com/$1$",
"path":"$.ADDRESS"
}
eol

Submitting the skill

The final part is adding the skill to the list of skills for SUSI. We can do it by 2 ways:

1st method (using the web interface)

  • Open https://susi.skills.com and login into SUSI account (or sign up, if not done).
  • Click on the create skill button.
  • Select the appropriate fields like Category, Language, Skill name, Logo.
  • Paste the text file that we had created.
  • Add comments regarding the skill and click on Save to save the skill.

2nd method (sending a PR)

  • Send a Pull Request to susi_skill_data repository providing the dream name. The PR should have the text file containing the skill.

So, this was a short blog on how we can develop a SUSI skill of our choice.

Resources

Implementing Skill Detail Section in SUSI Android App

SUSI Skills are rules that are defined in SUSI Skill Data repo which are basically the responses SUSI gives to the user queries. When a user queries something from the SUSI Android app, a query to SUSI Server is made which further fetches response from SUSI Skill Data and gives the response to the app. Similarly, when we need to list all skills, an API call is made to server to list all skills. The server then checks the SUSI Skill Data repo for the skills and then return all the required information to the app. Then the app displays all the information about the skill to user. User then can view details of each skill and then interact on the chat interface to use that skill. This process is similar to what SUSI Skill CMS does. The CMS is a skill wiki like interface to view all skills and then edit them. Though the app can not be currently used to edit the skills but it can be used to view them and try them on the chat interface.

API Information

For listing SUSI Skill groups, we have to call on /cms/getGroups.json

This will give you all groups in SUSI model in which skills are present. Current response:

{
  "session": {"identity": {
    "type": "host",
    "name": "14.139.194.24",
    "anonymous": true
  }},
  "accepted": true,
  "groups": [
    "Small Talk",
    "Entertainment",
    "Problem Solving",
    "Knowledge",
    "Assistants",
    "Shopping"
  ],
  "message": "Success: Fetched group list"
}

So, the groups object gives all the groups in which SUSI Skills are located.

Next comes, fetching of skills. For that the endpoint is /cms/getGroups.json?group=GROUP_NAME

Since we want all skills to be fetched, we call this api for every group. So, for example we will be calling http://api.susi.ai/cms/getSkillList.json?group=Entertainment for getting all skills in group “Entertainment”. Similarly for other groups as well.

Sample response of skill:

{
  "accepted": true,
  "model": "general",
  "group": "Shopping",
  "language": "en",
  "skills": {"amazon_shopping": {
    "image": "images/amazon_shopping.png",
    "author_url": "https://github.com/meriki",
    "examples": ["Buy a dress"],
    "developer_privacy_policy": null,
    "author": "Y S Ramya",
    "skill_name": "Shop At Amazon",
    "dynamic_content": true,
    "terms_of_use": null,
    "descriptions": "Searches items on Amazon.com for shopping",
    "skill_rating": null
  }},
  "message": "Success: Fetched skill list",
  "session": {"identity": {
    "type": "host",
    "name": "14.139.194.24",
    "anonymous": true
  }}
}

It gives all details about skills:

  1. image
  2. author_url
  3. examples
  4. developer_privacy_policy
  5. author
  6. skill_name
  7. dynamic_content
  8. terms_of_use
  9. descriptions
  10. skill_rating

Implementation in SUSI Android App

Skill Detail Section UI of Google Assistant

Skill Detail Section UI of SUSI SKill CMS

Skill Detail Section UI of SUSI Android App

The UI of skill detail section in SUSI Android App is the mixture of UI of Skill detail section in Google Assistant ap and SUSI Skill CMS. It displays details of skills in a beautiful manner with horizontal recyclerview used to display the examples.

So, we have to display following details about the skill in Skill Detail Section:

  1. Skill Name
  2. Author Name
  3. Skill Image
  4. Try it Button
  5. Description
  6. Examples
  7. Rating
  8. Content type (Dynamic/Static)
  9. Terms of Use
  10. Developer’s Privacy policy

Let’s see the implementation.

1. Whenever a skill Card View is clicked, showSkillDetailFragment() is called and it opens a new instance of a fragment named SkillDetailsFragment which shows details of the skill. We have to provide necessary information while starting the fragment. This information is passed as a Serializable.

fun showSkillDetailFragment(skillData: SkillData, skillGroup: String) {
   val skillDetailsFragment = SkillDetailsFragment.newInstance(skillData,skillGroup)
   (context as SkillsActivity).fragmentManager.beginTransaction()
           .replace(R.id.fragment_container, skillDetailsFragment)
           .commit()
}

2.  The data which was passed as a Serializeable object is now casted back to the required form and a method to set up the UI is called.

companion object {
   val SKILL_KEY = "skill_key"
   val SKILL_GROUP = "skill_group"
   fun newInstance(skillData: SkillData, skillGroup: String): SkillDetailsFragment {
       val fragment = SkillDetailsFragment()
       val bundle = Bundle()
       bundle.putSerializable(SKILL_KEY, skillData as Serializable)
       bundle.putString(SKILL_GROUP, skillGroup)
       fragment.arguments = bundle

       return fragment
   }
}

override fun onCreateView(inflater: LayoutInflater, container: ViewGroup?, savedInstanceState: Bundle?): View {
   skillData = arguments.getSerializable(
           SKILL_KEY) as SkillData
   skillGroup = arguments.getString(SKILL_GROUP)
   return inflater.inflate(R.layout.fragment_skill_details, container, false)
}

override fun onViewCreated(view: View?, savedInstanceState: Bundle?) {
   setupUI()
   super.onViewCreated(view, savedInstanceState)
}

3. The setupUI() method then calls separate method for setting every part of the UI like image, name etc.

fun setupUI() {
   setImage()
   setName()
   setAuthor()
   setTryButton()
   setDescription()
   setExamples()
   setRating()
   setDynamicContent()
   setPolicy()
   setTerms()
}

4. One example of setting a part of the UI is setting Author name. It checks if AuthorName is null or not. After that it anchors author’s github account link with his/her name.

fun setAuthor() {
   skill_detail_author.text = "Author : ${activity.getString(R.string.no_skill_author)}"
   if(skillData.author != null && !skillData.author.isEmpty()){
       if(skillData.authorUrl == null || skillData.authorUrl.isEmpty())
           skill_detail_author.text = "Author : ${skillData.skillName}"
       else {
           skill_detail_author.linksClickable = true
           skill_detail_author.movementMethod = LinkMovementMethod.getInstance()
           if (android.os.Build.VERSION.SDK_INT >= Build.VERSION_CODES.N) {
               skill_detail_author.text = Html.fromHtml("Author : <a href=\"${skillData.authorUrl}\">${skillData.author}</a>", Html.FROM_HTML_MODE_COMPACT)
           } else {
               skill_detail_author.text = Html.fromHtml("Author : <a href=\"${skillData.authorUrl}\">${skillData.author}</a>")
           }
       }
   }
}

Summary

So, this blog talked about how the Skill detail section in SUSI Android App is implemented. This included how a network call is made, logic for making different network calls, making a horizontal recyclerview for displaying examples. So, If you are looking forward to contribute to SUSI Android App, this can help you a little. But if not so, this may also help you in understanding and how you can implement horizontal recyclerview similar to Google Play Store.

References

  1. To know about servlets https://en.wikipedia.org/wiki/Java_servlet
  2. To see how to implement one https://www.javatpoint.com/servlet-tutorial
  3. To see how to make network calls in android using Retrofit https://guides.codepath.com/android/Consuming-APIs-with-Retrofit
  4. To see how to implement custom RecyclerView Adapter https://www.survivingwithandroid.com/2016/09/android-recyclerview-tutorial.html

Implementing Skill Listing in SUSI Android App using Nested RecyclerViews

SUSI Skills are rules that are defined in SUSI Skill Data repo which are basically the responses SUSI gives to the user queries. When a user queries something from the SUSI Android app, a query to SUSI Server is made which further fetches response from SUSI Skill Data and gives the response to the app. Similarly, when we need to list all skills, an API call is made to server to list all skills. The server then checks the SUSI Skill Data repo for the skills and then return all the required information to the app. Then the app displays all the information about the skill to user. User then can view details of each skill and then interact on the chat interface to use that skill. This process is similar to what SUSI Skill CMS does. The CMS is a skill wiki like interface to view all skills and then edit them. Though the app can not be currently used to edit the skills but it can be used to view them and try them on the chat interface.

API Information

For listing SUSI Skill groups, we have to call on  /cms/getGroups.json

This will give you all groups in SUSI model in which skills are present. Current response:

{
  "session": {"identity": {
    "type": "host",
    "name": "14.139.194.24",
    "anonymous": true
  }},
  "accepted": true,
  "groups": [
    "Small Talk",
    "Entertainment",
    "Problem Solving",
    "Knowledge",
    "Assistants",
    "Shopping"
  ],
  "message": "Success: Fetched group list"
}

So, the groups object gives all the groups in which SUSI Skills are located.

Next comes, fetching of skills. For that the endpoint is /cms/getGroups.json?group=GROUP_NAME

Since we want all skills to be fetched, we call this api for every group. So, for example we will be calling http://api.susi.ai/cms/getSkillList.json?group=Entertainment for getting all skills in group “Entertainment”. Similarly for other groups as well.

Sample response of skill:

{
  "accepted": true,
  "model": "general",
  "group": "Shopping",
  "language": "en",
  "skills": {"amazon_shopping": {
    "image": "images/amazon_shopping.png",
    "author_url": "https://github.com/meriki",
    "examples": ["Buy a dress"],
    "developer_privacy_policy": null,
    "author": "Y S Ramya",
    "skill_name": "Shop At Amazon",
    "dynamic_content": true,
    "terms_of_use": null,
    "descriptions": "Searches items on Amazon.com for shopping",
    "skill_rating": null
  }},
  "message": "Success: Fetched skill list",
  "session": {"identity": {
    "type": "host",
    "name": "14.139.194.24",
    "anonymous": true
  }}
}

It gives all details about skills:

  1. image
  2. author_url
  3. examples
  4. developer_privacy_policy
  5. author
  6. skill_name
  7. dynamic_content
  8. terms_of_use
  9. descriptions
  10. skill_rating

Implementation in SUSI Android App

Skill Listing UI of Google Assistant

Skill Listing UI of SUSI SKill CMS

Skill Listing UI of SUSI Android App

The UI of skill listing in SUSI Android App is the mixture of UI of Skill listing in Google Assistant ap and SUSI Skill CMS. It displays skills in a beautiful manner with horizontal recyclerview nested in vertical recyclerview.

So, for implementing horizontal recyclerview inside vertical recyclerview, you need two viewholders and two adapters (one each for a recyclerview).

Let’s see the implementation.

1. First task is to fetch the information of groups in which skills are located. This line calls method in SkillListModel which then makes an API call to fetch groups.

skillListingModel.fetchGroups(this)

2. When the API call is succeeded, the below mentioned method is called which then calls a  skillListingModel.fetchSkills(groups[0], this) which fetches the skills located in group[0] group.

override fun onGroupFetchSuccess(response: Response<ListGroupsResponse>) {
   if (response.isSuccessful && response.body() != null) {
       groupsCount = response.body().groups.size
       groups = response.body().groups
       skillListingModel.fetchSkills(groups[0], this)
   } else {
       skillListingView?.visibilityProgressBar(false)
       skillListingView?.displayErrorDialog()
   }
}

3. When API call for fetching skills in group[0] succeeds, the count value is increased and then skills in group[1] are fetched and so on.

override fun onSkillFetchSuccess(response: Response<ListSkillsResponse>, group: String) {
   if (response.isSuccessful && response.body() != null) {
       skills.add(Pair(group, response.body().skillMap))
       count++
       if(count == groupsCount) {
           skillListingView?.visibilityProgressBar(false)
           skillListingView?.updateAdapter(skills)
       } else {
           skillListingModel.fetchSkills(groups[count], this)
       }
   } else {
       skillListingView?.visibilityProgressBar(false)
       skillListingView?.displayErrorDialog()
   }
}

4. When skills in all groups are fetched, the data in adapter is updated using skillGroupAdapter.notifyDataSetChanged()

override fun updateAdapter(skills: ArrayList<Pair<String, Map<String, SkillData>>>) {
   this.skills.clear()
   this.skills.addAll(skills)
   skillGroupAdapter.notifyDataSetChanged()
}

5. The data is set to the layout in two adapters made earlier. The following is the code to set the group name and adapter to horizontal recyclerview. This is the GroupAdapter to set data to row item in vertical recyclerview.

override fun onBindViewHolder(holder: GroupViewHolder?, position: Int) {
   if(skills[position].first != null)
       holder?.groupName?.text = skills[position].first
   holder?.skillList?.setHasFixedSize(true)
   val mLayoutManager = LinearLayoutManager(context, LinearLayoutManager.HORIZONTAL, false)
   holder?.skillList?.layoutManager = mLayoutManager
   holder?.skillList?.adapter = SkillListAdapter(context, skills[position])
}

6. Similarly, the data of each individual element in the horizontal recyclerview is set in the skillAdapter. The data set are title, examples, description and image. We have used Picasso library to load images from the URL.

override fun onBindViewHolder(holder: SkillViewHolder?, position: Int) {
   val skillData = skillDetails.second.values.toTypedArray()[position]

   if(skillData.skillName == null || skillData.skillName.isEmpty()){
       holder?.skillPreviewTitle?.text = context.getString(R.string.no_skill_name)
   } else {
       holder?.skillPreviewTitle?.text = skillData.skillName
   }

   if( skillData.descriptions == null || skillData.descriptions.isEmpty()){
       holder?.skillPreviewDescription?.text = context.getString(R.string.no_skill_description)
   } else {
       holder?.skillPreviewDescription?.text = skillData.descriptions
   }

   if(skillData.examples == null || skillData.examples.isEmpty())
       holder?.skillPreviewExample?.text = StringBuilder("\"").append("\"")
   else
       holder?.skillPreviewExample?.text = StringBuilder("\"").append(skillData.examples[0]).append("\"")

   if(skillData.image == null || skillData.image.isEmpty()){
       holder?.previewImageView?.setImageResource(R.drawable.ic_susi)
   } else {
       Picasso.with(context.applicationContext).load(StringBuilder(imageLink)
               .append(skillDetails.first.replace(" ","%20")).append("/en/").append(skillData.image).toString())
               .fit().centerCrop()
               .into(holder?.previewImageView)
   }
}

Summary

So, this blog talked about how the Skill Listing feature in SUSI Android App is implemented. This included how a network call is made, logic for making different network calls, making a nested horizontal recyclerview inside vertical recyclerview, etc. So, If you are looking forward to contribute to SUSI Android App, this can help you a little. But if not so, this may also help you in understanding and how you can implement nested recyclerviews similar to Google Play Store.

References

  1. To know about servlets https://en.wikipedia.org/wiki/Java_servlet
  2. To see how to implement one https://www.javatpoint.com/servlet-tutorial
  3. To see how to make network calls in android using Retrofit https://guides.codepath.com/android/Consuming-APIs-with-Retrofit
  4. To see how to implement Horizontal recyclerView inside Vertical recyclerView http://android-pratap.blogspot.in/2015/12/horizontal-recyclerview-in-vertical.html
  5. To see how to implement custom RecyclerView Adapter https://www.survivingwithandroid.com/2016/09/android-recyclerview-tutorial.html

Uploading Images to SUSI Server

SUSI Skill CMS is a web app to create and modify SUSI Skills. It needs API Endpoints to function and SUSI Server makes it possible. In this blogpost, we will see how to add a servlet to SUSI Server to upload images and files.

The CreateSkillService.java file is the servlet which handles the process of creating new Skills. It requires different user roles to be implemented and hence it extends the AbstractAPIHandler.

Image upload is only possible via a POST request so we will first override the doPost method in this servlet.

  @Override
  protected void doPost(HttpServletRequest req, HttpServletResponse resp) throws ServletException, IOException {
  resp.setHeader("Access-Control-Allow-Origin", "*"); // enable CORS

resp.setHeader enables the CORS for the servlet. This is required as POST requests must have CORS enables from the server. This is an important security feature that is provided by the browser.

        Part file = req.getPart("image");
        if (file == null) {
            json.put("accepted", false);
            json.put("message", "Image not given");
        }

Image upload to servers is usually a Multipart Request. So we get the part which is named as “image” in the form data.

When we receive the image file, then we check if the image with the same name exists on the server or not.

Path p = Paths.get(language + File.separator + “images/” + image_name);

        if (image_name == null || Files.exists(p)) {
                json.put("accepted", false);
                json.put("message", "The Image name not given or Image with same name is already present ");
            }

If the same file is present on the server then we return an error to the user requesting to give a unique filename to upload.

Image image = ImageIO.read(filecontent);
BufferedImage bi = this.createResizedCopy(image, 512, 512, true);
if(!Files.exists(Paths.get(language.getPath() + File.separator + "images"))){
   new File(language.getPath() + File.separator + "images").mkdirs();
           }
ImageIO.write(bi, "jpg", new File(language.getPath() + File.separator + "images/" + image_name));

Then we read the content for the image in an Image object. Then we check if images directory exists or not. If there is no image directory in the skill path specified then create a folder named “images”.

We usually prefer square images at the Skill CMS. So we create a resized copy of the image of 512×512 dimensions and save that copy to the directory we created above.

BufferedImage createResizedCopy(Image originalImage, int scaledWidth, int scaledHeight, boolean preserveAlpha) {
        int imageType = preserveAlpha ? BufferedImage.TYPE_INT_RGB : BufferedImage.TYPE_INT_ARGB;
        BufferedImage scaledBI = new BufferedImage(scaledWidth, scaledHeight, imageType);
        Graphics2D g = scaledBI.createGraphics();
        if (preserveAlpha) {
            g.setComposite(AlphaComposite.Src);
        }
        g.drawImage(originalImage, 0, 0, scaledWidth, scaledHeight, null);
        g.dispose();
        return scaledBI;
    }

The function above is used to create a  resized copy of the image of specified dimensions. If the image was a PNG then it also preserves the transparency of the image while creating a copy.

Since the SUSI server follows an API centric approach, all servlets respond in JSON.

       resp.setContentType("application/json");
       resp.setCharacterEncoding("UTF-8");
       resp.getWriter().write(json.toString());’

At last, we set the character encoding and the character set of the output. This helps the clients to parse the data easily.

To see this endpoint in live send a POST request at http://api.susi.ai/cms/createSkill.json.

Resources

Apache Docs: https://commons.apache.org/proper/commons-fileupload/using.html

Multipart POST Request Tutorial: http://www.codejava.net/java-se/networking/upload-files-by-sending-multipart-request-programmatically

Java File Upload tutorial: https://ursaj.com/upload-files-in-java-with-servlet-api

Jetty Project: https://github.com/jetty-project/