👾Sentiment Endpoints

Data for sentimental analysis and decision making

Endpoints

Retrieves The Twitter Emotion Insights

GET https://api.wisdomise.io/v0/insights/twitter-emotions/

Emotion insights are divided into 6 classes which are as follows: sadness, joy, love, anger, fear, and surprise. The Sum of all classes is 1. We use a Bert-based neural network to classify the tweets.

Query Parameters

Name	Type	Description
date_time__lte*	String	to datetime in isoformat example: 2022-10-09T00:00:05 lte = less than equal to
date_time__gte*	String	from datetime in isoformat example: 2022-10-08T00:00:05 gte = greater than equal to
resolution	String	resolution Example: 1d Default: 1d
symbol	String	symbol pair Example: BTCUSDT If not provided, all symbol pair insights will be returned.

Headers

Name	Type	Description
Authorization*	String	Token <AUTH_TOKEN>

200: OK

{
    "count": 2,
    "next": null,
    "previous": null,
    "results": [
        {
            "id": 30,
            "key": "",
            "insight_service_name": "twitter-insight-emotion",
            "resolution": "1d",
            "received_at": "2022-10-24T18:57:10.549600Z",
            "created_at": "2022-10-24T18:57:10.419153Z",
            "date_time": "2022-10-24T18:00:00Z",
            "expiration_date_time": "2022-10-25T18:00:00Z",
            "market_property": "fundamental",
            "version": "twitter-insight_v0.1.2",
            "exchange": null,
            "symbol": "BTCUSDT",
            "insight_result": {
                "data": [
                    {
                        "datetime": "2022-10-24T18:00:00+00:00",
                        "joy": 0.5768059062933316,
                        "fear": 0.1425325441028302,
                        "love": 0.009024323328427272,
                        "anger": 0.17245961243833396,
                        "sadness": 0.09132212283138263,
                        "surprise": 0.00785548853870811
                    }
                ]
            }
        },
        ...
    ]
}

Retrieves The Twitter Sentiment Insights

GET https://api.wisdomise.io/v0/insights/twitter-sentiments/

Sentiment insights are divided into 3 classes which are as follows: positive, neutral, and negative. The Sum of all classes is 1. We use a Bert-based neural network to classify the tweets.

Query Parameters

Name	Type	Description
date_time__lte*	String	to datetime in isoformat example: 2022-10-09T00:00:05 lte = less than equal to
date_time__gte*	String	from datetime in isoformat example: 2022-10-08T00:00:05 gte = greater than equal to
resolution	String	resolution Example: 1d Default: 1d
symbol	String	symbol pair Example: BTCUSDT If not provided, all symbol pair insights are returned.

Headers

Name	Type	Description
Authorization*	String	Token <AUTH_TOKEN>

200: OK

{
    "count": 4,
    "next": null,
    "previous": null,
    "results": [
        {
            "id": 42,
            "key": "",
            "insight_service_name": "twitter-insight-sentiment",
            "resolution": "1d",
            "received_at": "2022-10-25T07:06:17.394298Z",
            "created_at": "2022-10-25T07:06:15.823680Z",
            "date_time": "2022-10-25T07:00:00Z",
            "expiration_date_time": "2022-10-26T07:00:00Z",
            "market_property": "fundamental",
            "version": "twitter-insight_v0.1.2",
            "exchange": null,
            "symbol": "ADAUSDT",
            "insight_result": {
                "data": [
                    {
                        "datetime": "2022-10-25T07:00:00+00:00",
                        "neutral": 0.5612352894675173,
                        "negative": 0.09630140576186126,
                        "positive": 0.34246330522841784
                    }
                ]
            }
        },
        ...
    ]
}

Retrieves The Twitter Bull/Bear Frequencies Insights

GET https://api.wisdomise.io/v0/insights/twitter-frequencies/

It clarifies the intensity of bullishness/bearishness of the last related tweets. We use bullish/bearish related keywords for frequency calculation. Please note that a tweet can be neither bullish nor bearish.

bull + bear <= 1

Query Parameters

Name	Type	Description
date_time__lte*	String	to datetime in isoformat example: 2022-10-09T00:00:05 lte = less than equal to
date_time__gte*	String	from datetime in isoformat example: 2022-10-08T00:00:05 gte = greater than equal to
resolution	String	resolution Example: 1d Default: 1d
symbol	String	symbol pair Example: BTCUSDT If not provided, all symbol pair insights are returned.

Headers

Name	Type	Description
Authorization*	String	Token <AUTH_TOKEN>

200: OK

{
    "count": 4,
    "next": null,
    "previous": null,
    "results": [
        {
            "id": 30,
            "key": "",
            "insight_service_name": "twitter-insight-bull-bear",
            "resolution": "1d",
            "received_at": "2022-10-24T18:55:39.272528Z",
            "created_at": "2022-10-24T18:55:39.148091Z",
            "date_time": "2022-10-24T18:00:00Z",
            "expiration_date_time": "2022-10-25T18:00:00Z",
            "market_property": "fundamental",
            "version": "twitter-insight_v0.1.2",
            "exchange": null,
            "symbol": "BTCUSDT",
            "insight_result": {
                "data": [
                    {
                        "datetime": "2022-10-24T18:00:00+00:00",
                        "bear": 0.3313333333333333,
                        "bull": 0.098,
                    }
                ]
            }
        },
        ...
    }
]

Word Connect (Available upon request)

GET https://api.wisdomise.io/v1/insights/sentiment/word-connect

Word Cloud (Available upon request)

GET https://api.wisdomise.io/v1/insights/sentiment/word-cloud

Keyword Trends (Available upon request)

GET https://api.wisdomise.io/v1/insights/sentiment/keyword-trend

Use Cases

Twitter

To understand the crypto market better, one must understand how different participants feel about various events; for instance, Elon Musk's tweets can have a significant effect on the market. There is no doubt that Twitter is the main battleground for different individuals regarding the status of the market. We can make better decisions with clean and clear insights from sentimental analysis.

As it is obvious, Twitter is the main battlefield of different people regarding the status of the market. Having clean and clarified insights from the sentimental analysis can help us with better decision-making.

All the mentioned models can be considered as input for another analyzer model or could be used as a gate for taking a decision.

Twitter Models Performance

• The F1 score for the sentiment classification model is about 72.9%.

• The F1 score for the emotion classification model is about 77.3 %.

• The Bull/Bear frequency analyzer model just considers related keywords for proportion calculation.

Word Connect

Indicates the relation between different nodes regarding the keywords being searched and queried around the web. By having this feature, overall sentiment of users around of the industry can be mined.

Word Cloud

Presents the frequency of usage of different words based on each coin. For illustration, you can find out that mining keyword is more popular for Bitcoin vs nft.

Keyword Trends

Understanding different users' tendencies based on the country and state can be really helpful for clustering web3 clients. In addition, it indicates the popularity of specific keywords as well.