The AI chat tool ChatGPT, which has recently continued to attract global attention, is a pre-trained language model developed by OpenAI, a Silicon Valley High Tech company. The model adopts the Transformer deep learning architecture proposed by Google in 2017 and uses a massive amount of language data for pre-training to achieve the ability to analyze grammar and semantic understanding of natural language. The model also includes some improvements to Transformer to adapt to specific issues in conversational tasks. The pre-training data comes from the internet, including a large amount of social media, news websites, Wikipedia, blogs, etc. Additionally, the multi-head attention mechanism introduced in Transformer can weigh and integrate long-range information in the context into the features of the current word, helping the model better understand various contexts in the conversation. Finally, the model uses reinforcement learning techniques to continually optimize the answer results by iteratively learning the best answer from multiple answers, so that the final published model can generate more accurate intelligent answers based on the user's text input. Since its official release in November 2022, ChatGPT has become the hottest investment theme for technology companies worldwide in 2023 due to its excellent performance in various language questions and answers. The financial industry is also actively exploring various use cases of ChatGPT.
Eikon Data API Introduction
LSEG's Eikon financial terminal is an open technology solution aimed at financial market professionals. Customers can not only view industry-leading data, insights, and exclusive credible news through the terminal, but also data scientists can directly access real-time data, historical data, news, and analysis from data sources such as exchanges, research institutions, governments, and news agencies through the Eikon Data API, provided by Eikon, and be able to seamlessly integrate with customer’s existing workflow using Python. The combination of Eikon Data API and Eikon Codebook provides a comprehensive and flexible platform for accessing financial data and analysis and has now become a popular choice for global financial institutions and other organizations that require high-quality financial data and analysis.
How to use ChatGPT to quickly learn the Eikon Data API
The Eikon Data API provides rich interfaces that help customers access and use high-quality global financial data in a programmatic way. Customers can obtain API user documentation from the company's website or join our well-developed developer community to find best practices. The emergence of ChatGPT provides our customers with a new and more efficient way to master the use of the Eikon Data API. As one of the largest financial market data and infrastructure providers in the world, the related API usage documentation and experience sharing of the Eikon Data API on the internet should have been adopted by ChatGPT's model training. Next, we will try to use ChatGPT to quickly learn how to use the Eikon Data API to obtain high-quality financial data and verify it with Eikon Codebook, hoping to help everyone.
- After logging into the official website of OpenAI, we first asked what preparation is needed before using the Eikon Data API with ChatGPT. We can see that ChatGPT's response is professional, friendly, detailed, and constructive.
- As the previous response mentioned the Developer Portal of Eikon Data API page, I'm asking the ChatGPT how can I access it
- Next, we asked ChatGPT a few questions about how to get basic stock information, stock industry classification, and historical prices. The Eikon Data API usage code generated by ChatGPT looks concise, clear, and professional.
- Due to the rich data fields provided in the Eikon terminal, to help users quickly browse the definitions and descriptions of these fields and to facilitate direct use in the Eikon Data API, we provide the Data Item Browser (DIB) tool in the terminal. We were pleasantly surprised to find that ChatGPT has also learned the usage guide of the Data Item Browser tool.
- Finally, we hope to verify that the Python code generated by ChatGPT can run directly on the Eikon platform and achieve the expected results. Therefore, we asked a question about generating a stock MACD curve in ChatGPT, producing the following code.
- We directly copy and paste the code above generated by ChatGPT into Eikon Codebook, with a minor change of ek.set_app_key, the code can draw the MACD curve of Apple Inc. as expected.
Summary and outlook
First, the deep learning-based artificial intelligence technology adopted by ChatGPT has brought revolutionary changes to our search and knowledge-learning experience. So, how can enterprises better utilize the language pre-training model to solve their own natural language processing business pain points? One way is through the open sharing of the corpus, that is, putting more high-quality documents and corpus that can be publicly disclosed on the Internet and establishing a well-interactive customer community, so that this information can be more easily included in the next model training by ChatGPT, helping ChatGPT model to constantly improve and optimize the answer effect in more scenarios. When you and your colleagues are sincerely and candidly discussing problems in elegant natural language, perhaps you have not perceived that Confucius's "knowing is knowing, not knowing is not knowing" is unconsciously affecting your current conversation, but ChatGPT can, through the attention mechanism, discover how these unconscious factors affect our conversation and help us constantly improve the level of conversation.
Second, ChatGPT is just the starting point of a new AI revolution, and there is still huge room for development in the industry in the future. ChatGPT mainly solves natural language processing problems, and currently only handles language text. With the development of artificial intelligence in multimodal pre-training models, etc., will there be a multimodal version of ChatGPT in the future that supports multiple modes such as language, image, and voice while interacting at the same time? For example, you can generate related financial analysis charts according to one language description or generate technical analysis proposals according to images of a set of K-line charts.
However, if you have any questions regarding our API usage and you would like to get helps from our experts other than the ChatGPT, please feel free to check and participate in our Q&A Forum. We're looking forward to participate with you there!