How to teach SUSI skills calling an External API

SUSI is an intelligent  personal assistant. SUSI can learn skills to understand and respond to user queries better. A skill is taught using rules. Writing rules is an easy task and one doesn’t need any programming background too. Anyone can start contributing. Check out these tutorials and do watch this video to get started and start teaching susi.

SUSI can be taught to call external API’s to answer user queries.

While writing skills we first mention string patterns to match the user’s query and then tell SUSI what to do with the matched pattern. The pattern matching is similar to regular expressions and we can also retrieve the matched parameters using $<parameter number>$ notation.

Example :

My name is *
Hi $1$!

When the user inputs “My name is Uday” , it is matched with “My name is *” and “Uday” is stored in $1$. So the output given is “Hi Uday!”.

SUSI can call an external API to reply to user query. An API endpoint or url when called must return a JSON or JSONP response for SUSI to be able to parse the response and retrieve the answer.

Rule Format for a skill calling an external API

The rule format for calling an external API is :

<regular expression for pattern matching>
!console: <return answer using $object$ or $required_key$>
{
  “url”:<API endpoint or url>,
  “path”:$.<key in the API response to find the answer>,
}
eol
  • Url is the API endpoint to be called which returns a JSON or JSONP response.
    The parameters to the url if any can be added using $$ notation.
  • Path is used to help susi know where to look for the answer in the returned response.
    If the path points to a root element, then the answer is stored in $object$, otherwise we can query $key$ to get the answer which is a value to the key under the path.
  • eol or end of line indicates the end of the rule.

Understanding the Path Attribute

Let us understand the Path attribute better through some test cases.

In each of the test cases we discuss what the path should be and how to retrieve the answer for a given required answer from the json response of an API.

  1. API response in json :

    {
      “Key1” : “Value1”
    }
    

Required answer : Value1
Path : “$.Key1    =>   Retrieve Answer:  $object$

 

  1. API response in json :
{
  “Key1” : [{“Key11” : “Value11”}]
}

Required answer : Value11
Path : $.Key1[0]   =>  Retrieve Answer: $Key11$
Path : $.Key1[0].Key11   => Retrieve Answer: $object$

 

  1. API response in json :
{
  “Key1” : {“Key11” : “Value11”}
}


Required answer : Value11
Path : $.Key1  => Retrieve Answer:  $Key11$
Path : $.Key1.Key11  => Retrieve Answer: $object$

 

  1. API response in json :
{
  “Key1” : {
           “Key11” : “Value11”,
           “Key12” : “Value12”
         }
}

Required answer : Value11 , Value12
Path : $.Key1  => Retrieve Answer:  $Key11$ , $Key12$

Where to write these rules?

Now, since we know how to write rules let’s see where to write them.

We use etherpads to write and test rules and once we finish testing our rule we can push those rules to the repo.

Steps to open, write and test rules:

  1. Open a new etherpad with a desired name <etherpad name> at http://dream.susi.ai/
  2. Write your skills code in the etherpad following the code format explained above.
  3. Now, to test your skill let’s chat with susi. Start a conversation with susi at http://susi.ai/chat to test your skills.
  4. Load your skills by typing dream <etherpad name> and wait for a response saying dreaming enabled for <etherpad name>
  5. Test your skill and follow step 4 every time you make changes to the code in your etherpad.
  6. After you are done testing, type stop dreaming and if you are satisfied with your skill do send a PR to help susi learn.

Examples

Let us try an example to understand this better.

1. Plot of a TV Show

Tvmaze is an open  TV API that provides information about tv shows. Let us write a rule to know the plot of a tv show. We can find many such APIs. Check out this link listing few of them.

  1.  Open an etherpad at http://dream.susi.ai/ named tvshowplot.
  2.   Enter the code to query plot of a TV show in the etherpad at                           http://dream.susi.ai/p/tvshowplot
* plot of *|* summary of *
!console:$object$
{
  "url":"http://api.tvmaze.com/singlesearch/shows?q=$2$",
  "path":"$.summary"
}
eol
  1. Now lets test our skill by starting a conversation with susi at http://susi.ai/chat.
  • User Query: dream tvshowplot
    Response:  dreaming enabled  for tvshowplot
  • User Query: what is the plot of legion
    Response: Legion introduces the story of David Haller: Since he was a teenager, David has struggled with mental illness. Diagnosed as schizophrenic, David has been in and out of psychiatric hospitals for years. But after a strange encounter with a fellow patient, he’s confronted with the possibility that the voices he hears and the visions he sees might be real. He’s based on the Marvel comics character Legion, the son of X-Men founder Charles Xavier (played by Patrick Stewart and James McAvoy in the films), first introduced in 1985.

Intermediate Processing:

Pattern Matching : $1$ = “what is the” ; $2$ = “legion”

Url : http://api.tvmaze.com/singlesearch/shows?q=legion

API response:

{
  "id": 6393,
  "url": "http:\/\/www.tvmaze.com\/shows\/6393\/legion",
  "name": "Legion",
  "type": "Scripted",
  "language": "English",
  "genres": [
             "Drama",
             "Action",
             "Science-Fiction"
            ],
  "summary": "<p><strong>Legion<\/strong> introduces the story of David Haller: Since he was a teenager, David has struggled with mental illness. Diagnosed as schizophrenic, David has been in and out of psychiatric hospitals for years. But after a strange encounter with a fellow patient, he's confronted with the possibility that the voices he hears and the visions he sees might be real. He's based on the Marvel comics character Legion, the son of X-Men founder Charles Xavier (played by Patrick Stewart and James McAvoy in the films), first introduced in 1985.<\/p>",
  "updated": 1491955072,
}

Note: The API response has been trimmed to show only the relevant content.

Path : $.summary

Retrieving Answer: so our required answer in the api response is under the key summary and is retrieved using $object$ since it is a root element.

 

Screenshots:

2. Cooking Recipes

Let us try it out with another API.
Recipepuppy is an cooking recipe API where users can query various recipes.

  1.  Open a etherpad at http://dream.susi.ai/ named recipe.
  2.   Enter the code to query a recipe in the etherpad at  http://dream.susi.ai/p/recipe
#Gives recipes and links to cook a dish
* cook *
!console:<p>To cook  <strong>$title$</strong> : <br>The ingridients required are: $ingredients$. <br> For instruction to prepare the dish $href$ </p>
{
  "url":"http://www.recipepuppy.com/api/?q=$2$",
  "path":"$.results"
}
eol
  1. Now lets test our skill by starting a conversation with susi at http://susi.ai/chat.
  • User Query: dream recipe
    Response:  dreaming enabled  for recipe
  • User Query: how to cook chicken biryani
    Response: To cook Chicken Biryani Recipe :
    The ingridients required are: chicken, seeds, chicken broth, rice, butter, peas, garlic, red onions, cardamom, curry paste, olive oil, tomato, coriander, cumin, brown sugar, tumeric.
    For instruction to prepare the dish Click Here!

Intermediate Processing:

Pattern Matching : $1$ = “how to” ; $2$ = “chicken biryani”

Url : http://www.recipepuppy.com/api/?q=chicken biryani

API response:

{
  "title": "Recipe Puppy",
  "version": 0.1,
  "href": "http:\/\/www.recipepuppy.com\/",
  "results": [
              {
                "title": "Chicken Biryani Recipe",
                "href": "http:\/\/www.grouprecipes.com\/53040\/chicken-biryani.html",
                "ingredients": "chicken, seeds, chicken broth, rice, butter, peas, garlic, red onions, cardamom, curry paste, olive oil, tomato, coriander, cumin, brown sugar, tumeric",
                "thumbnail": "http:\/\/img.recipepuppy.com\/413822.jpg"
             },
           ]
}

Note: The API response has been trimmed to show only the relevant content.

Path : $.results[0]

Retrieving Answer: so our required answer in the api response is under the key results and since it’s an array we are using the first element of the array and since the element is a dictionary too we use its keys correspondingly to answer. The $href$ is rendered as “Click Here” hyperlinked to the actual url.

 

Screenshots:

 

We have successfully taught susi a skill which tells users about the plot of a tv show and a skill to query recipes.
Cheers!
Following similar procedure, we can make use of other APIs and teach susi several new skills.

Resources
Continue Reading

Deploying Susi Server on Google Cloud with Kubernetes

Susi (acronym for Scientific User Support Intelligence) is an advanced AI made by people at FOSSASIA. It is an AI made by the people and for the people.
Susi is an Open Source Project under LGPL Licence.

SUSI.AI already has many Skills and anyone can add new skills through simple console rules.

If you want to participate in the development of the SUSI server you can start by learning to deploy it on a cloud system like Google Cloud.

This way whenever you make a change to Susi Server, you can test it out on various Susi Apps instantly.

Google Cloud with Kubernetes provide this ability. Let’s dig deep into what is Google Cloud Platform and Kubernetes.

What is Google Cloud Platform ?

Google Cloud Platform lets you build and host applications and websites, store data, and analyze data on Google’s scalable infrastructure.
Google Cloud Platform (at the time of writing this article) also provides free credits worth $300 for 1 year for testing out the Platform and test your applications.

What is Kubernetes ?

Kubernetes is an open-source system for automatic deployment, management and scaling of containerized applications. It makes it easy to roll out updates to your application with simple commands from your development machine and scale horizontally easily by adding more clusters as demand increase.

Deploying Susi Server on Kubernetes

Deploying Susi Server on Kubernetes is a fairly easy task. Follow up the steps to get it running.

Create a Google Cloud Account

Sign up for a Google Cloud Account (https://cloud.google.com/free-trial/) and get 300$ credits for initial use.

Create a New Project

After successful sign up, create a new project on Google Cloud Console.
Let’s name it Susi-Kubernetes . 

You will be provided a ProjectID. Remember it for further reference.

Install Google Cloud SDK and kubectl

Go to https://cloud.google.com/sdk/ and see instructions to setup Google Cloud SDK on your respective OS.

After Google Cloud SDK install, run

gcloud components install kubectl

This will install kubectl for interacting with Kubernetes.

Login and setup project

  1. Login to your Google Cloud Account using
$ gcloud auth login

2. List all the projects using

$ gcloud config list project
[core]
project = <PROJECT_ID>

3. Select your project

$ gcloud config set project <PROJECT_ID>

4. Install JDK8 for susi_server setup and set it as default.

5. Clone your fork of the Susi Server Repository

$ git clone https://github.com/<your_username>/susi_server.git
$ cd susi_server/

6. Build project and run Susi Server locally

$ ./gradlew build
$ bin/start.sh

Susi server must have been started started and web interface is accessible on http://localhost:4000

Install Docker and build Docker image for Susi

  1. Install Docker.
    Debian and derivatives:  sudo apt install docker
    Arch Linux:   sudo pacman -S docker 
  2. Build Docker Image for Susi
    $ docker build -t gcr.io/<Project_id>/susi:v1 .
  3. Push Image to Google Container Registry private to your project.
$ gcloud docker -- push gcr.io/<Project_id>/susi:v1

Create Cluster and Deploy your Susi Server there

  1. Create Cluster. You may specify different zone, number of nodes and machine type depending upon requirement.
    $ gcloud container clusters create <Cluster-Name> --num-nodes 2 --machine-type n1-standard-1 --zone us-central1-c
  2. Run your deployment. You may specify any name for deployment.
    $ kubectl run <deployment_name> --image=gcr.io/<Project_id>/susi:v1 --port=80
    $ kubectl get deployments
    $ kubectl expose deployment susi --type=LoadBalancer
  3. Check your deployment and get Public IP for Access.
    $ kubectl get services
    NAME         CLUSTER-IP     EXTERNAL-IP     PORT(S)       AGE
    kubernetes   10.3.240.1     <none>          443/TCP        1d
    susi         10.3.241.145   <PUBLIC_IP>     80:31155/TCP   1d
  4. Go to provided public IP to check, if Susi Server is running.

Congratulations, you successfully setup Susi Server on Google Cloud with Kubernetes.

Updating the deployment

Next step is to update deployment when you wish to roll out changes. To do so.

Build Docker Image and Push it to Google Container Registry

$ docker build -t gcr.io/<Project_Id>/susi:v2 .
$ gcloud docker -- push gcr.io/<Project_Id>/susi:v2

Update Deployment Image with Kubernetes

$ kubectl set image deployment/<Deployment_Name> \
  <Deployment_Name>=gcr.io/<Project_id>/susi:v2
deployment "<Deployment_Name>" image updated

Go to public ip to see the changes.

That’s it. Now, you have fully running Susi Server on your own Google Cloud Cluster using Kubernetes.

Continue Reading

Susi AI Skill Development

What is Susi?

Susi is an open source intelligent personal assistant which has the capability to learn and respond better to queries. It is also capable of making to-do lists, setting alarms, providing weather and traffic info all in real time. Susi responds based on skills.

What is a skill? How do we teach a skill?

A skill is a piece of code which performs a set of actions in order to respond to the user’s query. These skills are based on pattern matching which help them mapping the user’s query to a specific skill and responding accordingly. Teaching a skill to Susi is surprisingly very easy to implement. One can take a look at the Susi Skill Development Tutorial and a video workshop by Michael Christen.

I will try to give a basic idea on how to create a skill, it’s basic structure and some of the skills I developed in the first week.

Prepare to create a skill:

  • Head over to http://dream.susi.ai
  • Create a etherpad with some relevant name
  • Delete all text currently present in there
  • Start writing your skill

Adding to this, for testing a skill one can head over to Susi Web Chat Interface.

Basic Structure for calling an API:

<Regular expression to be matched here>

!console:<response given to the user>
 {
 "url":"<API endpoint>",
 "path":"<Json path here>"
 }
 eol

So, let me explain this line by line.

  1. The regular expression is the one to which the user’s query is matched first.
  2. The console is meant to output the actual response the user sees as response.
  3. In place of the “url”, the API endpoint is passed in.
  4. “path” here specifies how we traverse through the response Json or Jsonp to get the object, starts with “$.”.
  5. At last, “eol” which is the end-of-line marks the end of a skill.

Let’s take an example for better understanding of this:

random gif
!console: $url$
{
    "url" : "http://api.giphy.com/v1/gifs/trending?api_key=dc6zaTOxFJmzC",
    "path" : "$.data[0].images.fixed_height"
}
eol 

 

This skill responds with a link to a random gif.

Steps involved:

  1. Match the string “random gif” with the user’s query.
  2. On successful match, make an API call to the API endpoint specified in “url”
  3. On response, extract the object at the specified path in the json under “path”
  4. Respond to the user with the “url” key’s value which would here be an URL of a GIF.

Let’s try it out on Susi Web Chat. For this, you will first have to load your skill using the dream command followed by etherpad name: dream <etherpad name>. And then you can start testing your skill.

So, we queried “random gif” and we got a response “Click Here!”. The complete URL didn’t show up because all the URLs are currently parsed and a hyperlink for each is created. So try clicking on it to find a GIF.

 

Now, let’s look at one more skill I developed during this period.

# Returns the name of the president of a country

 president of *|who is the president of *| president *
 !console:$plaintext$
 {      "url":"https://api.wolframalpha.com/v2/query?input=president+$1$&output=JSON&appid=9WA6XR-26EWTGEVTE&includepodid=Result",
   "path" : "$.queryresult.pods[0].subpods[0]"
 }
 eol

 

Let’s understand this step by step:

  1. We have here “president of *|who is the president of *| president *”, which means the user’s query matches with anyone of the following because of the use of pipe symbol “|”. The “*” here replaces a word or a list of words, which can be accessed like: “${index}$”  where index is replaced by the position of the “*” in the expression starting from 1.
  2. Now we have something new in the URL. See that  $1$  inside the URL? On runtime, that is replaced with the content of the “*” variable. So if a user puts in query like: “president of usa”, “usa” is mapped to $1$ and is replaced in the URL and appropriate API request is made.
  3. Then the path is traversed in the json response and the value of the “plaintext” key is used to respond to the user.

 

It’s now time to try it out on Susi Web Chat.

So, we got our desired response here, i.e., the name of the president of usa.

Continue Reading

Generating a documentation site from markup documents with Sphinx and Pandoc

Generating a fully fledged website from a set of markup documents is no easy feat. But thanks to the wonderful tool sphinx, it certainly makes the task easier. Sphinx does the heavy lifting of generating a website with built in javascript based search. But sometimes it’s not enough.

This week we were faced with two issues related to documentation generation on loklak_server and susi_server. First let me give you some context. Now sphinx requires an index.rst file within /docs/  which it uses to generate the first page of the site. A very obvious way to fill it which helps us avoid unnecessary duplication is to use the include directive of reStructuredText to include the README file from the root of the repository.

This leads to the following two problems:

  • Include directive can only properly include a reStructuredText, not a markdown document. Given a markdown document, it tries to parse the markdown as  reStructuredText which leads to errors.
  • Any relative links in README break when it is included in another folder.

To fix the first issue, I used pypandoc, a thin wrapper around Pandoc. Pandoc is a wonderful command line tool which allows us to convert documents from one markup format to another. From the official Pandoc website itself,

If you need to convert files from one markup format into another, pandoc is your swiss-army knife.

pypandoc requires a working installation of Pandoc, which can be downloaded and installed automatically using a single line of code.

pypandoc.download_pandoc()

This gives us a cross-platform way to download pandoc without worrying about the current platform. Now, pypandoc leaves the installer in the current working directory after download, which is fine locally, but creates a problem when run on remote systems like Travis. The installer could get committed accidently to the repository. To solve this, I had to take a look at source code for pypandoc and call an internal method, which pypandoc basically uses to set the name of the installer. I use that method to find out the name of the file and then delete it after installation is over. This is one of many benefits of open-source projects. Had pypandoc not been open source, I would not have been able to do that.

url = pypandoc.pandoc_download._get_pandoc_urls()[0][pf]
filename = url.split(‘/’)[-1]
os.remove(filename)

Here pf is the current platform which can be one of ‘win32’, ‘linux’, or ‘darwin’.

Now let’s take a look at our second issue. To solve that, I used regular expressions to capture any relative links. Capturing links were easy. All links in reStructuredText are in the same following format.

`Title <url>`__

Similarly links in markdown are in the following format

[Title](url)

Regular expressions were the perfect candidate to solve this. To detect which links was relative and need to be fixed, I checked which links start with the \docs\ directory and then all I had to do was remove the \docs prefix from those links.

A note about loklak and susi server project

Loklak is a server application which is able to collect messages from various sources, including twitter.

SUSI AI is an intelligent Open Source personal assistant. It is capable of chat and voice interaction and by using APIs to perform actions such as music playback, making to-do lists, setting alarms, streaming podcasts, playing audiobooks, and providing weather, traffic, and other real time information

Continue Reading
Close Menu