CODEWITHSUNDEEP
node.jsmachine-learningchatbotwatson-discoverywatson-assistant
mahmoudbustami
mahmoudbustami

Reputation: 35

i can't receive any answer from Watson Discovery

I am working in a chatbot with Watson Assistant and Discovery but I can't receive any answer from Discovery this is the dialog when I used the discovery intent enter image description here

and this is the output from the UI that I designed enter image description here

function updateMessage(res, input, response) {
// if (!response.output) {
//     response.output = {};
// } else 
if (response.output.action === 'callDiscovery') {
  // if you want to use natural_language_query, set here the input from the user like my example: 
  params.natural_language_query = response.input.text || null;

    console.log('Calling discovery');
    discovery.query(params, (error, returnDiscovery) => {
        if (error) {
          next(error);
        } else {
          console.log('return from discovery: '+returnDiscovery);
          //if you want to send all text returned from discovery, discomment these lines
            var text;
              for (i = 0; i < returnDiscovery.results.length; i++) {
                text += returnDiscovery.results[i].text + "<br>";
              }
          //sending the TEXT returned from discovery results
          response.output.text = 'Discovery call with success, check the results: <br>' + text; //results */
          //sending the PASSAGES returned from discovery results
          response.output.text = 'Discovery call with success, check the results: <br>' + returnDiscovery.passages[0].passage_text; //passageResults
          return res.json(response);
        };
    });
} else if (response.output && response.output.text) {
    return res.json(response);
}

}

Upvotes: 0

Views: 196

Answers (2)

MNA
MNA

Reputation: 323

The tokenizer argument in the TfidfVectorizer is for overriding the string tokenization step. For e.g. you can use a function as shown below which accepts string as argument, tokenizes the string and return tokenized words. 

def tokenizerFunc(x):
    return x.split()

This function accepts string as input and returns list of words. The reason you get the error "init() takes 1 positional argument but 2 were given" is because function 'WordNetLemmatizer()' does not accept any argument, however when using this function within TfidfVectorizer string is passed as an argument for it to be tokenized.

In case you want to lemmatize and tokenze simultaneously you can use this function below 

lemmatizer = WordNetLemmatizer()
def tokenizerFunc(x):
    tokenizedList = x.split()
    lemmatizedList = [ lemmatizer.lemmatize(i) for i in tokenizedList] 
    return lemmatizedList

you have to use it like this 

TFIDF = TfidfVectorizer(tokenizer=tokenizerFunc,analyzer= 'word',min_df=3,token_pattern=r'(?u)\b[A-Za-z]+\b',stop_words= 'english') 
tfidf_matrix = TFIDF.fit_transform(df2['job_title'])

Upvotes: 1

qaiser
qaiser

Reputation: 2868

tokenizer should be lemmatizer.lemmatize and not lemmatizer

 lemmatizer = WordNetLemmatizer()
 TFIDF = TfidfVectorizer(tokenizer=lemmatizer.lemmatize,analyzer= 'word',min_df=3,token_pattern=r'(?u)\b[A-Za-z]+\b',stop_words= 'english')

output

TFIDF.fit_transform(['how are you', 'facing issue','hope this well help you' ])
#o/p
<3x3 sparse matrix of type '<class 'numpy.float64'>'
with 9 stored elements in Compressed Sparse Row format>

Upvotes: 2

Related Questions