How to Generate ONIX Keywords with ChatGPT: Using AI in Your Publishing Workflow

ONIX Article Header

At our recent event on practical tips for incorporating AI into your publishing workflow, Alison Coles, Data Manager & Programmer at Boydell & Brewer, shared her experience of using generative AI tools, such as the free version of ChatGPT, or similar AI tool, Claude, to create ONIX keywords.

Dive into Alison’s valuable insights and pick up tips for embracing the iterative nature of working with AI, safely using data that’s already in the public domain (ONIX metadata or website data, for example) – including how you can hone effective prompting strategies to improve your results – in this step-by-step guide.

Relax: a trial and error approach is OK!

Even with programming experience, working with AI involves a lot of trial and error. No one should expect to get it right the first time, and neither should you expect AI to produce perfect output immediately. Think of the process as a learning journey between the user and the AI and remember that it’s all about iteration to hone the outcomes you’re presented with.

Conversation is key: engaging with AI gets results

When prompting AI tools such as ChatGPT, it can start with a simple question and quickly become a deep dive, as AI might not give you exactly what you want first time. The key is to be patient, have fun, and not be afraid to ask questions to build on the results in ways that make sense for your goals. This approach helps in refining the prompts and improving the AI’s responses.

Crafting an effective prompt

Alison has learned through her time experimenting that an effective prompt should contain the following elements:

  1. Data Source: the raw information you feed into the AI. Be careful if using the free version of ChatGPT that you aren’t inputting any sensitive, private or copyrighted material. For Alison’s project, she was able to use publicly available details of the titles in question.
  2. Context: provides background on how to approach the task.
  3. Task Instructions: include specific directions on what you want the AI to do.
  4. Output Instructions: specify clear guidelines on the format of the expected output.

Example prompt for generating ONIX keywords

Jump into this practical example that Alison shared of a prompt used to generate keywords with ChatGPT. 

1. Data Source 

The data is sourced from an Excel spreadsheet containing ISBN, title, and description. You can simply paste this data into ChatGPT, which handles Excel data well.

2. Additional Data 

To improve keyword generation, additional data like table of contents, reviews, and author biographies can be included. This data is usually publicly available and non-copyrighted.

3. Context

The first line of the prompt sets the context:

Context
You are a person searching the internet for books to expand your thirst for knowledge.

TIP: This helps AI understand the perspective from which it should generate keywords.

4. Task Instructions

Task Instructions
Generate ONIX keywords for each of the books in the complete provided text. Be as specific as possible and choose words that would maximise their discoverability in search engines.

TIP: Using “generate” immediately prompts the AI to produce output. Specifying “complete provided text” ensures the AI reads all the data and not just a part of it.

5. Output Instructions

Output Instructions
Output the list in CSV format, the order of the item should be ISBN, title, keywords, separate the keywords with a semicolon.

TIP: This specifies the exact format for the output, making it easy to integrate the results into existing systems.

Who’s steering the car? Why AI needs good drivers

For general prompting, Alison also recommends asking, in one sentence, for AI to outline a presentation or generate a training document on a certain subject. The results can be extremely useful, but in all cases, she finds that the best results always come from the human who understands the task and the desired outcome. Humans will know enough about the job to recognise when the AI’s output is incorrect and this expertise ensures that the AI is used effectively and efficiently.

Boosting and supercharging your work

Alison’s key piece of advice is to experiment with the provided prompt, tailor it to your needs, and use it as inspiration for creating your own prompts. 

By focusing on the data source, providing clear context, and giving precise instructions, you can harness the power of AI to enhance your publishing workflow. It’s not just about automating tasks, but about using AI as a tool that complements your expertise, making your work more efficient and effective. What’s not to love?


Want to access a recording of our AI event? The full video, Using AI to Supercharge Your Publishing Work(flows), is available via the BookMachine Campus content library for our subscribers. Find out more about joining the community to catch up with the recording, and gain access to 100s of industry resources, perks and discounts.

Related Articles

Sign up to our Newsletter

Subscribe

* indicates required

BookMachine Ltd. will use the information you provide on this form to be in touch with you and to provide updates and marketing. Please let us know all the ways you would like to hear from us:

You can change your mind at any time by clicking the unsubscribe link in the footer of any email you receive from us, or by contacting us at hello@bookmachine.org. We will treat your information with respect. For more information about our privacy practices please visit our website. By clicking below, you agree that we may process your information in accordance with these terms.

We use Mailchimp as our marketing platform. By clicking below to subscribe, you acknowledge that your information will be transferred to Mailchimp for processing. Learn more about Mailchimp’s privacy practices.