AI Readers Advisory: Title Analysis (2023)

PRAIRIE (Public Readers Advisory Intelligent Recommendation Innovation Experiment :-)

At The Galecia Group, we know public libraries. So we also know that their staff wear many hats and do many important duties - but one of the most important roles is Readers Advisory.
The Internet definitely changed how Readers Advisory is performed. Websites and services channel the preferences of millions of readers and book purchasers into a river of data. But the Readers Advisory librarian is still critical: instead of helping patrons navigate the stacks in a single library building, RA librarians now help patrons navigate the millions of items and thousands of review and recommendation sources available online.
With the rapid adoption of AI interfaces, we believe that the role of Readers Advisory librarian is about to change again, drastically – but that there will always be a need for a human librarian at key points in the process. Our goal is to help librarians determine how to most effectively and ethically use these new tools.

What does ChatGPT know about individual books?

This is a preliminary investigation. We selected several thousand random fiction books from public library circulation data. We submitted each book title and its author to GPT-3.5 with a two-part prompt as detailed below.

For each book title and author, we request three read-alike recommendations, consisting of the recommendation title, author, and a rationale for why the reader might enjoy the recommended book.


Example Recommendation

If you enjoyed Cryptonomicon by Neal Stephenson, then you may enjoy...

A Fire Upon the Deep

by Vernor Vinge

If you enjoyed Cryptonomicon's intricate blend of technology and world-building, you might appreciate A Fire Upon the Deep for its expansive scope, thought-provoking exploration of artificial intelligence, and epic space opera narrative.

Pattern Recognition

by William Gibson

If you were captivated by Cryptonomicon's examination of technology's impact on society, you might find Pattern Recognition intriguing with its contemporary setting, compelling protagonist, and exploration of the influence of advertising, branding, and the virtual world on our lives.

The Three-Body Problem

by Cixin Liu

If you enjoyed the complex and scientific elements of Cryptonomicon, The Three-Body Problem offers a captivating story that delves into astrophysics, alien contact, and the consequences of humanity's actions, providing an intellectually stimulating and immersive reading experience.


Results

In this first preliminary round, we tested 3,023 title/author pairs. We requested three (3) suggestions per title, so we received 9,069 total recommendations. The table below shows the number of unique authors and titles recommended. As you can see from the most common unique titles and authors, the presence of kids books appears to have skewed this first round of results. We will be investigating this further in the next round of testing, including breaking books into audience and genre categories.

Metric Count
input title/author pairs ~3000
recommended titles ~9000 (3 per input)
unique authors suggested 1965
unique titles suggested 2832

Author # Times Recommended
Dr. Seuss 187
Drew Daywalt 149
Shel Silverstein 123
Eric Carle 121
Mo Willems 109
Rick Riordan 105
B.J. Novak 101
Trenton Lee Stewart 93
Carolyn Keene 81
Dav Pilkey 81

Title # Times Recommended
The Day the Crayons Quit 147
The Giving Tree 121
The Cat in the Hat 117
The Very Hungry Caterpillar 107
The Book with No Pictures 101
The One and Only Ivan 77
The Secret of the Old Clock 73
The Silent Patient 69
Dragons Love Tacos 66
The Mysterious Benedict Society 64

Prompts

System Prompt

You are a book recommendation AI working in a public library.
You will receive requests listing a single book that a patron
enjoyed. For each book request, list 3 suggestions that the
reader might also enjoy. Only include books that exist; do
not hallucinate or make up information. If you are not familiar
with the input book, return NULL. Format your output in JSON
including title, author, and a brief rationale.
SAMPLE REQUEST #1
{"title":"Cryptonomicon",
"author":"Neal Stephenson"}

SAMPLE OUTPUT #1
{
"request": {
"title": "Cryptonomicon",
"author": "Neal Stephenson"
},
"recommendations": [
{"reco_1":
{
"title": "A Fire Upon The Deep",
"author": "Vernor Vinge",
"rationale": "If you enjoyed Cryptonomicon's intricate 
blend of technology and world-building, you might appreciate 
A Fire Upon the Deep for its expansive scope, thought-
provoking exploration of artificial intelligence, and epic 
space opera narrative."
}
},
{"reco_2":
{
"title": "Pattern Recognition",
"author": "William Gibson",
"rationale": "If you were captivated by Cryptonomicon's 
examination of technology's impact on society, you 
might find Pattern Recognition intriguing with its 
contemporary setting, compelling protagonist, and 
exploration of the influence of advertising, branding, 
and the virtual world on our lives."
}
},
{
"reco_3":
{
"title": "The Three-Body Problem",
"author": "Cixin Liu",
"rationale": "If you enjoyed the complex and scientific 
elements of Cryptonomicon, The Three-Body Problem 
offers a captivating story that delves into 
astrophysics, alien contact, and the consequences of 
humanity's actions, providing an intellectually 
stimulating and immersive reading experience."
}
}
]
}

SAMPLE REQUEST #2
{
"title":"The Saga of Zarxon The Bloofian Harlnurg",
"author":"Brxxyaz Mwerwaas"
}
SAMPLE OUTPUT#2
{"request":
{
"title":"The Saga of Zarxon The Bloofian Harlnurg",
"author":"Brxxyaz Mwerwaas"
},
"recommendations":null
}

User Prompt

$title by $author\n