We say we live in an era where data is the new oil, but we have limited power over this massive body of data. Moreover, this data, most of the time, stays as raw data without making something useful out of it. Businesses are continuously seeking new tools to extract practical insights from the vast sea of unstructured data available on the internet. One of the substantial sources of this data is social media, which hosts a plethora of user-generated content that delivers a goldmine of information about customer preferences, demands, and brand standing. Nevertheless, harnessing this data productively requires sophisticated tools capable of processing and interpreting human language. It is where artificial intelligence comes in, with technologies like Keyword Extraction API, Sentiment Analysis API, and Large Language Models like ChatGPT leading the charge. Here in this tutorial, you will learn how to create a ChatGPT app using a combination of other APIs.
Table of Contents
How Does ChatGPT App Gauge the Emotional Tone in Sentiment Analysis?
It’s crucial to understand how individuals feel about what they’re talking about. This is where the Sentiment Analysis API from APILayer comes in. This API can analyze text input using natural language processing to determine if it is good, negative, or neutral.
However, the Sentiment Analysis API’s capabilities extend beyond simple sentiment analysis. It also includes capabilities such as automated customer service, text categorization, language identification, and chatbot installation.
Moreover, in addition to that, extracting keywords from the sentence/context can be helpful for further analysis. For example, the Keyword Extraction API from APILayer is an effective approach for identifying important themes for content marketing, analyzing client comments about your goods, or getting ideas for new products or services.
This API helps organizations find key subjects in a sea of unstructured data, assisting in the identification of trends, monitoring brand reputation, and understanding consumer sentiment. It may also assist in the generation of tags for blog posts, the translation of papers into other languages, and the summarization of huge materials.
What is Sentiment Analysis, exactly?
Sentiment analysis is a branch of natural language processing that detects data sentiment. Companies regularly use sentiment analysis to explore genuine client demands and analyze company reputation by analyzing various sorts of texts, such as customer reviews, social media chats, support requests, and so on. Sentiment analysis is used to determine information polarity (positive vs. negative), emotion (anger, happiness, sadness, etc.), and intention (e.g., interested vs. not interested).
Who Makes Extensive Use of Sentiment Analysis in Industry?
Sentiment analysis, often known as opinion mining, identifies and extracts subjective information from source materials using natural language processing, text analysis, and computational linguistics.
Banks and other service providers are increasingly relying on artificial intelligence-powered sentiment analysis to assess client emotions, attitudes, and views in real time. For example, these tools may measure the general attitude towards a specific product, service, or brand by analyzing social media postings, customer reviews, and other user-generated information. This data may then be utilized to evaluate strengths and shortcomings in their offers, follow changes in consumer sentiment over time, and more effectively respond to customer issues. Here you can get your free API access key, our powerful Sentiment Analysis API!
How Can the ChatGPT App Help in Creating Reports and Extracting Insights?
Businesses may use the ChatGPT app to provide insights, reports, and recommended actions after extracting important keywords and assessing the emotion of the talks. ChatGPT, an OpenAI AI language model, can read and create human-like prose, making it a great tool for data analysis.
For example, after performing keyword extraction and sentiment analysis on social media data, ChatGPT may create a report summarising the main themes of conversation, the general sentiment towards these topics, and any noteworthy changes in sentiment over time. Here we have a demo application for you where we showcase how to extract keywords and do a sentiment analysis using APILayer APIs. Then we can send all the analyses to the OpenAI ChatGPT API endpoint to analyze and get insights.
The first step is to get the API access key from APILayer for Keyword Extraction and Sentiment Analysis API along with the OpenAI ChatGPT API key. Then you can download the latest version of Delphi Community Edition, where we can create multi-platform native applications using Delphi and FireMonkey.
As you can see in the above image, our ChatGPT app is being built using the FireMonkey GUI framework, and we are using RESTClient components to work with APILayer APIs and to connect with OpenAI ChatGPT API – we will utilize TNetHTTPClient components to connect with the endpoint. Since we are dedicating this post to creating the ChatGPT app, you might find some functions of this demo are not completed, except the main functions like:
- Check Keywords
- Sentiment Analysis
- Processing them with ChatGPT API
Here is the Keyword Extraction process:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 |
procedure TFormMain.BtnCheckKeywordsClick(Sender: TObject); begin RESTClient1.ResetToDefaults; RESTClient1.Accept := 'application/json, text/plain; q=0.9, text/html;q=0.8,'; RESTClient1.AcceptCharset := 'utf-8, *;q=0.8'; RESTClient1.BaseURL := 'https://api.apilayer.com/keyword'; RESTClient1.ContentType := 'application/json'; RESTRequest1.ClearBody; RESTRequest1.Params.Clear; // API acc. key RESTRequest1.Params.AddItem; RESTRequest1.Params.Items[0].Kind := pkHTTPHEADER; RESTRequest1.Params.Items[0].Name := 'apikey'; RESTRequest1.Params.Items[0].Value := APILAYER_API_KEY; RESTRequest1.Params.Items[0].Options := [poDoNotEncode]; // set the url for checking RESTRequest1.Params.AddItem; RESTRequest1.Params.Items[1].Kind := pkREQUESTBODY; RESTRequest1.Params.Items[1].Name := 'body'; RESTRequest1.Params.Items[1].Value := MemoContent.Text; // sample body inserted RESTRequest1.Params.Items[1].ContentTypeStr := CONTENTTYPE_TEXT_PLAIN; RESTRequest1.Execute; // keyword analysis result MemoResult.Lines.Add(RESTResponse1.Content); FKeywords := RESTResponse1.Content; end; |
This one is the Sentiment Analysis API Endpoint integration:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
procedure TFormMain.BtnSentimentAnalysisClick(Sender: TObject); var LClient: TNetHTTPClient; LRequest: TNetHTTPRequest; LResponse: IHTTPResponse; LBody: TStringStream; begin LClient := TNetHTTPClient.Create(nil); LRequest := TNetHTTPRequest.Create(nil); try LRequest.Client := LClient; LRequest.MethodString := 'POST'; LRequest.URL := 'https://api.apilayer.com/sentiment/analysis'; LRequest.CustomHeaders['apikey'] := APILAYER_API_KEY; // Assuming that you have your content as string LBody := TStringStream.Create('body text content comes here', TEncoding.UTF8); try LResponse := LRequest.Post(LRequest.URL, LBody); if LResponse.StatusCode = 200 then begin // ShowMessage('Request was successful.') MemoResult.Lines.Add(LResponse.ContentAsString(TEncoding.UTF8)); FSentiment := LResponse.ContentAsString(TEncoding.UTF8) end else ShowMessage('Request failed. Status: ' + LResponse.StatusCode.ToString); finally LBody.Free; end; finally LRequest.Free; LClient.Free; end; end; |
Here is the explanation for these code samples. If you understand one, you will see little difference in the second.
TFormMain.BtnCheckKeywordsClick(Sender: TObject), is associated with a button’s OnClick event on a GUI form, meaning it is triggered when the button (presumably labeled “Check Keywords”) is clicked by a user.
Here’s a brief rundown of what this function does:
- It first resets the RESTClient1 object to its default settings. This object is responsible for making HTTP requests to RESTful services.
- It then sets up the REST client’s properties. The Accept property is set to prefer JSON, with text/plain and text/html as fallbacks. AcceptCharset is set to prefer UTF-8, and the BaseURL for the requests is set to ‘https://api.apilayer.com/keyword’, which is presumably an API for keyword analysis. The ContentType is set to ‘application/json’, meaning the request will send data in JSON format.
- It clears the body and parameters of RESTRequest1, preparing it for the new request.
- It adds two parameters to RESTRequest1. The first is the HTTP header ‘apikey’, presumably used for authentication with the API, with a value stored in APILAYER_API_KEY. The second parameter is a request body named ‘body’, which contains the text from MemoContent.Text (presumably a text input field in the GUI). This text is sent with the content type ‘text/plain’.
- The request is executed with the Execute method.
- Once the response is received, it’s added to the MemoResult.Lines (presumably a result display field in the GUI) and stored in the FKeywords variable. This response presumably contains the results of the keyword analysis performed by the API.
In summary, this function is used to perform a keyword analysis on the content of a text input field using the API provided by ‘api.apilayer.com’, and then it displays the results in the GUI and stores them in a variable.
How to Integrate OpenAI ChatGPT API in Delphi?
Now let’s see how to integrate OpenAI ChatGPT API and create insights using our keyword and sentiment analysis data:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
procedure TFormMain.BtnProcessClick(Sender: TObject); var Prompt: String; LClient: TNetHTTPClient; LRequest: TNetHTTPRequest; LResponse: IHTTPResponse; LBody: TStringStream; LJsonBody: TJSONObject; begin // create a prompt template Prompt := Format ('The text: { %s } was analyzed. Here is the keyword analysis result: { %s }. And Sentiment analysis result: { %s }. What insights can we draw from this complete text using these keywords and sentiment analysis', [MemoContent.Text, FKeywords, FSentiment]); LClient := TNetHTTPClient.Create(nil); LRequest := TNetHTTPRequest.Create(nil); try LRequest.Client := LClient; LRequest.MethodString := 'POST'; LRequest.URL := 'https://api.openai.com/v1/completions'; LRequest.Accept := 'application/json'; LRequest.CustomHeaders['Content-Type'] := 'application/json'; LRequest.CustomHeaders['Authorization'] := 'Bearer ' + OPENAI_API_KEY; LJsonBody := TJSONObject.Create; try LJsonBody.AddPair('model', 'text-davinci-003'); LJsonBody.AddPair('prompt', Prompt + '\n\nQ:'); LJsonBody.AddPair('temperature', TJSONNumber.Create(0)); LJsonBody.AddPair('max_tokens', TJSONNumber.Create(100)); LJsonBody.AddPair('top_p', TJSONNumber.Create(1)); LJsonBody.AddPair('frequency_penalty', TJSONNumber.Create(0)); LJsonBody.AddPair('presence_penalty', TJSONNumber.Create(0)); LJsonBody.AddPair('stop', TJSONArray.Create(TJSONString.Create('\n'))); LBody := TStringStream.Create(LJsonBody.ToString, TEncoding.UTF8); try LResponse := LRequest.Post(LRequest.URL, LBody); if LResponse.StatusCode = 200 then begin ShowMessage('Request was successful.'); MemoResult.Lines.Add(LResponse.ContentAsString(TEncoding.UTF8)); end else ShowMessage('Request failed. Status: ' + LResponse.StatusCode.ToString); finally LBody.Free; end; finally LJsonBody.Free; end; finally LRequest.Free; LClient.Free; end; end; |
While this is using preconfigured prompt, you can always change the prompt and get better and extended results. For instance, you can change a bit, requesting that I need various short posts related to this analyzed text with similar keywords and sentiment levels. That would give you various useful content that you can use on your social media accounts.
Here is our application in action. Besides, get the complete source code on this GitHub repository. Here in the screenshot, you can see that our app is showing the insights in the bottom text field:
|
1 |
From the keyword and sentiment analysis, we can draw the insight that the text is describing a peaceful and magical environment. The gentle breeze and whispering secrets suggest a tranquil atmosphere, while the mention of a magical world implies a sense of wonder and awe. The sentiment analysis result of \"neutral\" further confirms this impression. |
Moreover, check out our latest articles related to AI:
- Create Amazing Projects With These Free APIs in 2023
- What Is Open API? Pros, Cons, and Examples
- What Is API Integration: A Comprehensive Guide for Non-Technical People
Seizing the Future: The Potential of Artificial Intelligence in Social Media Monitoring
Businesses may construct a complete social media monitoring system by combining the capability of APILayer’s Keyword Extraction API and Sentiment Analysis API with the language comprehension capabilities of ChatGPT. This system has the potential to automate the process of filtering through social media data, finding major subjects of debate, analyzing public mood, and producing actionable insights.
Using these tools improves productivity and allows for a degree of understanding that would be difficult, if not impossible, to reach by human analysis alone. As organizations traverse an increasingly digital world, techniques such as keyword extraction, sentiment analysis, and AI language models will become increasingly important for keeping ahead of the competition.
Stay ahead of the competition and harness the power of artificial intelligence in social media monitoring. Explore the capabilities of APILayer’s Keyword Extraction API, Sentiment Analysis API, and OpenAI’s ChatGPT API by visiting the APILayer platform and OpenAI API page.
Frequently Asked Questions (FAQs)
How Does Sentiment Analysis API Work?
The API sorts the sentiment into three categories: positive, negative, and neutral. This lets businesses find out what customers think, rate their brand’s image, and make decisions based on the sentiment analysis results.
How Keyword Extraction Finds Relevant Keywords?
The API uses natural language processing methods to analyze the text and pull out important keywords. This gives useful information for apps like content marketing, sentiment analysis, and finding trends.
How To Get More Insightful Output?
In our example, we used a simple prompt to give a simple idea. By rewriting the prompt, you can achieve different results. For instance, you can ask ChatGPT API to create 10 posts related to that context with the exact keywords and sentiment levels.
