IssueTracker

Problem you have encountered:

This is regarding Analyzing Sentiment in Cloud Natural Language API tested English and Japanese texts.

For example, English text of “I won the gateball tournament.” shows the result of  {"magnitude": 0, "score": 0} and Japanese text of “ゲートボール大会で優勝した” and “ゲートボール大会で優勝した !!” shows {"magnitude": 0, "score": 0} and  {"magnitude": 0.1, "score": -0.1 }.

Suppose to say, there should be a positive meaning but, it can be also possible to learn “gateball” as a negative meaning in other side. Or It may be unclear in this respect.
If it is available, I want to know to what extent it categorize words for emotional analysis as much information as possible.

Steps to reproduce:

[Test Text : request.json]
{ "document":{"type":"PLAIN_TEXT", "content":"I won the gateball tournament." },"encodingType": "UTF8"}
>“I won the gateball tournament.” and “ゲートボール大会で優勝した”
>“I won the gateball tournament !!" and “ゲートボール大会で優勝した !!"

[Execute]
curl "https://language.googleapis.com/v1/documents:analyzeSentiment?key=${XXXXXX}"   -s -X POST -H "Content-Type: application/json" --data-binary @request.json

[Test Result]
>“I won the gateball tournament.” and “ゲートボール大会で優勝した”
"documentSentiment": {"magnitude": 0, "score": 0 },..."sentiment": {"magnitude": 0,"score": 0}

>“ゲートボール大会で優勝した !!"
"documentSentiment": {"magnitude": 0.1, "score": -0.1 },..."sentiment": {"magnitude": 0.1,"score": -0.1}

>“I won the gateball tournament !!"
"documentSentiment": {"magnitude": 0.1, "score": 0.1 },..."sentiment": {"magnitude": 0.1,"score": 0.1}


Other information (workarounds you have tried, documentation consulted, etc):

Referring following documents.
[1] https://cloud.google.com/natural-language/docs/basics
[2] https://cloud.google.com/natural-language/docs/analyzing-sentiment
[3] https://cloud.google.com/natural-language/docs/morphology