Scraping knowledge from webpages is a comparatively superior job that, till lately, required a level of technical ability. The thought of diving into code or scripts for knowledge extraction appeared overwhelming for a lot of, myself included.
Information scraping can energy many search engine optimization duties, comparable to auditing, competitor evaluation, and analyzing web site and knowledge construction.
Google sheets presents easy options to assist.
A type of options is the IMPORTXML perform that permits customers to scrape webpage knowledge utilizing only a few parameters. It makes knowledge extraction accessible to a wider viewers, particularly to those that weren’t well-versed in programming languages.
Whereas this perform is spectacular, the actual breakthrough got here with the adoption and integration of generative AI into the combination.
On this information, we’ll present you the right way to use Google Sheets and AI, significantly ChatGPT, for net scraping without having superior coding expertise.
The Instruments: AI And Chatbots
We at the moment are all conversant in AI, ChatGPT, and related chatbots.
The truth is, many people use options like ChatGPT to write down our personal code, scripts, and applications with or without very restricted programming data.
It is so simple as offering detailed directions within the type of prompts and dealing with the chatbot to construct instruments that solely till lately we believed have been approach above us.
However most significantly, these are instruments which might be deeply altering the best way we strategy our day-to-day work.
For instance, if we ask ChatGPT the next query, “What’s the IMPORTXML perform and the way can I exploit it in Google Sheets to scrape the title of an HTML webpage? Present the required code to do this in Google Sheets,” the response is extraordinarily correct. In a matter of seconds, we’ve our system prepared to make use of in Google Sheets.
However to be trustworthy, that was a really fundamental and easy job that we may have simply accomplished with out ChatGPT.
The Activity
So, how does this work if we wish to extract knowledge that may be a bit much less commonplace in comparison with a web page title or description?
For instance, how does this work if we wish to extract the next knowledge from the PPC entrance web page of Search Engine Journal?
Record all featured articles, their authors, the hyperlink URLs, and the article description for the columns listed on https://www.searchenginejournal.com/class/paid-media/pay-per-click/.
Can we do this immediately with ChatGPT?
Executing With ChatGPT
When creating prompts, it took just a few makes an attempt to offer directions that have been detailed sufficient for the chatbot to completely perceive the target of the duty and return good outcomes.
In lots of instances, it felt just like the AI was beneath strain to return fast outcomes regardless of their accuracy.
However let me clarify.
The duty was to investigate the web page and checklist all featured articles, their authors, the hyperlink URLs, and the outline for every of the 30 articles listed on the web page. Then compile the info right into a desk and eventually export it right into a CSV file.
Easy proper?
At first, ChatGPT returned only a pattern of seven articles and solely their titles and URLs; after a reworked immediate, it managed to checklist and export all 30 articles and their hyperlinks.
Now, that was good. So, to finish the duty, we simply wanted so as to add the authors and the article descriptions.
However right here is the place the bot stumbled and was not in a position to present an correct description of every article regardless of us offering examples of the web page ingredient it wanted to seek out and duplicate.
ChatGPT saved ignoring the directions and offering its personal article descriptions time and time once more.
ChatGPT even failed once we tried with a special strategy and downloaded and uploaded a replica of the web page HTML.
Screenshot from ChatGPT, February 2024
This time, it was in a position to present correct knowledge for seven articles however couldn’t go previous that. The difficulty reported:
“…the construction and content material of the web page current vital challenges for complete knowledge extraction in a single session.
The web page is kind of intensive and complicated, and it’s not possible to extract all 30 articles within the present format of interplay.”
Screenshot from ChatGPT, February 2024
ChatGPT + Google Sheets
So, going again to IMPORTXML and Google Sheets.
This time, getting ChatGPT to offer the formulation for every area was like a breeze.
Screenshot from ChatGPT, February 2024
Listed below are among the formulation, as recommended by the chatbot, which you could simply strive your self in Google Sheets to extract:
Title
=IMPORTXML(“https://www.searchenginejournal.com/class/paid-media/pay-per-click/”, “//*[@id=’archives-wrapper’]/article/div/div[2]/h2/a”)
Writer Identify
=IMPORTXML(“https://www.searchenginejournal.com/class/paid-media/pay-per-click/”, “//*[@id=’archives-wrapper’]/article/div/div[2]/p[1]/a”)
URL Hyperlink
=IMPORTXML(“https://www.searchenginejournal.com/class/paid-media/pay-per-click/”, “//*[@id=’archives-wrapper’]/article/div/div[2]/h2/a/@href”)
Description
=IMPORTXML(“https://www.searchenginejournal.com/class/paid-media/pay-per-click/”, “//*[@id=’archives-wrapper’]/article/div/div[2]/p[2]”)
Very quickly, we have been in a position to extract the info into the spreadsheet.
Screenshot from Google Sheets, February 2024
Moreover, through the use of merely constructed nested formulation, we will shortly pull the info from a number of pages on the identical time.
Within the instance under, I used to be in a position to extract the identical knowledge associated to every article (title, creator, URL hyperlink, and outline) for the primary 10 pages of the PPC part.
The result’s a complete of 300 articles scraped in lower than a minute!
Screenshot from Google Sheets, February 2024
Evaluating The Two
So, how do ChatGPT vs. ChatGPT + Google Sheets IMPORTXML evaluate?
In my expertise, I couldn’t discover a simple and fast approach to make use of ChatGPT to scrape the info I used to be in search of – thoughts you, that doesn’t imply that this isn’t attainable, and there could be a number of methods to do that, however I didn’t discover any.
What labored for me was a mix of the completely different instruments, and that served me rather well for my supposed goal.
ChatGPT was extraordinarily helpful for writing the IMPORTXML formulation I wanted to make use of in Google Sheets, and people formulation did the remainder.
An extra bonus of the ChatGPT + Google Sheets possibility is which you could simply use the free 3.5 model of ChatGPT and get the instrument to construct your IMPORTXML formulation, as a substitute of getting model 4 to scan the web page and extract the info.
Key Takeaway
This highlights a essential side of how AI has reworked how we expect and work.
The very best instrument for the job isn’t merely utilizing AI, Google Sheets, or any particular software program alone however somewhat a mix of instruments and expertise.
It’s on this built-in strategy that we develop workflows which might be environment friendly and efficient, thus bettering our general productiveness.
Extra sources:
Featured Picture: Visible Era/Shutterstock