Help With Making A ChatBot

SalladShooter · August 7, 2023, 11:59am

Question:
So I am making a simple chatbot named SaladChat. I keep getting this error →

Error: Invalid or missing intents in corpus.json 
repl process died unexpectedly: exit stays 1

Repl link:
https://replit.com/@SalladShooter/SaladChat#

JavaScript:

const fs = require('fs');
const readline = require('readline-sync');
const natural = require('natural');

const tokenizer = new natural.WordTokenizer();
const stemmer = natural.PorterStemmer;

let corpus;

try {
  const corpusData = fs.readFileSync('./corpus.json', 'utf8');
  corpus = JSON.parse(corpusData);
} catch (error) {
  console.error('Error reading or parsing corpus.json:', error.message);
  process.exit(1);
}

if (!corpus.intents || !Array.isArray(corpus.intents)) {
  console.error('Error: Invalid or missing intents in corpus.json');
  process.exit(1);
}

const intents = corpus.intents.map(intent => ({
  tag: intent.tag,
  patterns: intent.patterns.map(pattern => tokenizer.tokenize(pattern.toLowerCase())),
  responses: intent.responses
}));

function preprocessInput(input) {
  const tokens = tokenizer.tokenize(input.toLowerCase());
  return tokens.map(token => stemmer.stem(token));
}

function matchIntent(input) {
  const processedInput = preprocessInput(input);
  for (const intent of intents) {
    for (const pattern of intent.patterns) {
      const stemmedPattern = pattern.map(token => stemmer.stem(token));
      const similarity = natural.JaroWinklerDistance(
        processedInput.join(' '),
        stemmedPattern.join(' ')
      );
      if (similarity > 0.8) {
        return intent;
      }
    }
  }
  return null;
}

function chat() {
  const userInput = readline.question('You: ');
  if (userInput.toLowerCase() === 'exit') {
    console.log('Chatbot: Goodbye!');
    process.exit(0);
  }
  const matchedIntent = matchIntent(userInput);
  if (matchedIntent) {
    const randomResponse =
      matchedIntent.responses[Math.floor(Math.random() * matchedIntent.responses.length)];
    console.log('Chatbot:', randomResponse);
  } else {
    console.log("Chatbot: I'm sorry, I don't understand.");
  }
  chat();
}

console.log('Chatbot: Hello! How can I assist you today? (Type "exit" to quit)');
chat();

corpus.json:

[
  {
    "tag": "greeting",
    "patterns": ["hello", "hi", "hey"],
    "responses": ["Hello there!", "Hi!", "Hey!"]
  },
  {
    "tag": "goodbye",
    "patterns": ["bye", "goodbye", "see you"],
    "responses": ["Goodbye!", "Take care!", "See you later!"]
  }
]

WindLother · August 7, 2023, 12:50pm

Shouldn’t your corpus.json include intents?
Like this:

{
  "intents": [
    {
      "tag": "greeting",
      "patterns": ["hello", "hi", "hey"],
      "responses": ["Hello there!", "Hi!", "Hey!"]
    },
    {
      "tag": "goodbye",
      "patterns": ["bye", "goodbye", "see you"],
      "responses": ["Goodbye!", "Take care!", "See you later!"]
    }
  ]
}

SalladShooter · August 7, 2023, 12:55pm

@WindLother thanks that worked but I now get this error:

WindLother · August 7, 2023, 1:02pm

You need to know which value is undefined, can’t tell just looking at the error.

I’d suggest adding a console.log in matchIdent function.

function matchIntent(input) {
  const processedInput = preprocessInput(input);
  for (const intent of intents) {
    for (const pattern of intent.patterns) {
      const stemmedPattern = pattern.map(token => stemmer.stem(token));
      console.log('processedInput:', processedInput); // Add this line
      console.log('stemmedPattern:', stemmedPattern); // Add this line
      const similarity = natural.JaroWinklerDistance(
        processedInput.join(' '),
        stemmedPattern.join(' ')
      );
      if (similarity > 0.8) {
        return intent;
      }
    }
  }
  return null;
}

SalladShooter · August 7, 2023, 3:45pm

@WindLother this is the error I get:

WindLother · August 7, 2023, 5:17pm

Try to loop through the individual words and compare each.

For example instead of:

const similarity = natural.JaroWinklerDistance(
    processedInput.join(' '),
    stemmedPattern.join(' ')
);

Try:

for (let i = 0; i < processedInput.length; i++) {
    for (let j = 0; j < stemmedPattern.length; j++) {
        const similarity = natural.JaroWinklerDistance(processedInput[i], stemmedPattern[j]);
        if (similarity > 0.8) {
            return intent;
        }
    }
}

The variables are arrys but you need to pass strings to the function.

SalladShooter · August 7, 2023, 6:48pm

@WindLother I get this error now:

WindLother · August 7, 2023, 6:57pm

Before we proceed did you install the natural library?

npm install natural

SalladShooter · August 7, 2023, 7:36pm

@WindLother yes I have ran that in the shell.

UMARismyname · August 7, 2023, 7:48pm

looks like you need to have a third argument to JaroWinklerDistance, an options argument. If you hover over the function name in the IDE with code intelligence enabled, it should give you the function’s signature. I think you want to have this argument as { ignoreCase: true }

SalladShooter · August 8, 2023, 12:09pm

@UMARismyname where would I put that in my code?

UMARismyname · August 8, 2023, 1:20pm

Not sure what your code looks like, so CtrlH, click “enable Regular Expressions” in the top right, and replace (?<=JaroWinklerDistance\([^,]*,[^,]*)(?=\)) with { ignoreCase: true }

SalladShooter · August 8, 2023, 1:24pm

@UMARismyname Ctrl+H just backspaces like delete. Cmd+H just goes to my Home Screen (I’m on iPad btw).

UMARismyname · August 8, 2023, 1:47pm

oh, idk how to get to replace on ipad. So add , { ignoreCase: true } before the closing ) every time you call JaroWinklerDistance(

SalladShooter · August 9, 2023, 11:51am

@UMARismyname I get this error now →

This is my code now →

const fs = require('fs');
const readline = require('readline-sync');
const natural = require('natural');

const tokenizer = new natural.WordTokenizer();
const stemmer = natural.PorterStemmer;

let corpus;

try {
  const corpusData = fs.readFileSync('./corpus.json', 'utf8');
  corpus = JSON.parse(corpusData);
} catch (error) {
  console.error('Error reading or parsing corpus.json:', error.message);
  process.exit(1);
}

if (!corpus.intents || !Array.isArray(corpus.intents)) {
  console.error('Error: Invalid or missing intents in corpus.json');
  process.exit(1);
}

const intents = corpus.intents.map(intent => ({
  tag: intent.tag,
  patterns: intent.patterns.map(pattern => tokenizer.tokenize(pattern.toLowerCase())),
  responses: intent.responses
}));

function preprocessInput(input) {
  const tokens = tokenizer.tokenize(input.toLowerCase());
  return tokens.map(token => stemmer.stem(token));
}

function matchIntent(input) {
  const processedInput = preprocessInput(input);
  for (const intent of intents) {
    for (const pattern of intent.patterns) {
      const stemmedPattern = pattern.map(token => stemmer.stem(token));
      console.log('processedInput:', processedInput); // Add this line
      console.log('stemmedPattern:', stemmedPattern); // Add this line
      for (let i = 0; i < processedInput.length; i++) {
        for (let j = 0; j < stemmedPattern.length; j++) {
          const similarity = natural.JaroWinklerDistance(processedInput[i], stemmedPattern[j], { ignoreCase: true });
          if (similarity > 0.8) {
            return intent;
          }
        }
      }
      if (similarity > 0.8) {
        return intent;
      }
    }
  }
  return null;
}

function chat() {
  const userInput = readline.question('You: ');
  if (userInput.toLowerCase() === 'exit') {
    console.log('Chatbot: Goodbye!');
    process.exit(0);
  }
  const matchedIntent = matchIntent(userInput);
  if (matchedIntent) {
    const randomResponse =
      matchedIntent.responses[Math.floor(Math.random() * matchedIntent.responses.length)];
    console.log('Chatbot:', randomResponse);
  } else {
    console.log("Chatbot: I'm sorry, I don't understand.");
  }
  chat();
}

console.log('Chatbot: Hello! How can I assist you today? (Type "exit" to quit)');
chat();

WindLother · August 9, 2023, 12:32pm

You’re trying to access the variable similarity outside of the nested loops where it was defined.

Edit: showing in your code becausee I did a sloppy job answering

for (let i = 0; i < processedInput.length; i++) {
    for (let j = 0; j < stemmedPattern.length; j++) {
        const similarity = natural.JaroWinklerDistance(processedInput[i], stemmedPattern[j], { ignoreCase: true });
        if (similarity > 0.8) {
            return intent;
        }
    }
}
if (similarity > 0.8) {   // <---- You are trying to access this outside of the nested loop
    return intent;
}

When you do if (similarity > 0.8) check outside of the loops, the program say that this is redundant since the inner check (within the loops) already does the same thing.

Let just the similarity inside the nested loop and that should solve the problem.