A web-based public library software solution for advice on books to read, by comparing your list of excellent books against similar lists (Profiles). Your list, in turn, becomes part of the knowledge base so Advisor can serve the next user even better. Virtually all the "labour" is provided by users.
Negotiations are progressing with two regional libraries to trial the system. The goal is to increase the value readers get from the books they choose, leading to more rewarding reading, and a cumulative benefit for humanity.
I am sure I have missed some wonderful books. I am frequently surprised in second-hand bookshops. And I want a way of being informed of books which appeal to people like me.
The problem is that I don't know what I want, but I am sure it is out there, and some readers who like the same kind of books as me would know about it. But how to communicate through so much diversity?
A web server that asks for a list of books "you would recommend to someone like yourself", your favourites. Call this your Profile. The server goes to each book in your profile and follows links to every other profile which also suggests it. The suggested books on these other profiles are then ranked and the first page of Dorrigo Advisor's recommendations sent back to you. Your profile can also remember books you do not want to be told about again.
The OOP design has objects for books and profiles, cross-linked (many-to-many in DB jargon). The knowledge base is memory resident for performance (a flat file on disk for save/restore). A gigagyte of RAM is easily and cheaply available to handle lots of profiles and large collections. Sharing of profiles anonymously between Advisor sites is expected.
The web server is based on a simple Socket component listening to port 80. There is no Apache and cgi interpreter overhead. The package is written in Delphi on Windows platforms, but will be ported to Linux as soon as Kylix is released. The main engine is only a few hundred lines of Pascal code, loops within loops. The hard part was providing fast lookup of title/author and searching for a match to creative spelling in user entry. Details of all books in the regional catalogue is required for performance of this task.
Privacy should not be an issue, but provision is made for using p.i.n. with Barcode to log in. Your profile is lightly encrypted in saved knowledge base on disk, but the only identifying data it holds is your barcode which requires the main library system to translate. Anonymous login is allowed, with continuation of prior session. I have found that reading and writing HTML to provide a sweet browser interface is a skill that takes time to master.
Are you a regional library keen to be a pilot site? Two are in discussion stage but moving slowly.
The Dorrigo Advisor server connects to the LAN at the regional server farm. It is dependent on the library software for knowledge about local holdings, locations, and to streamline online reservation. The knowledge base will refer to books outside the regional collection, enriching cross-references and advice. It will favour local holdings in presenting suggestions.
Library staff may educate borrowers and direct them to Advisor. The algorithm does away with staff effort. Users add to the knowledge base in the process of asking for advice. Users' spelling is authenticated with catalogue (or more widely as a background task). Very bad spelling can be reviewed by administrator or ignored.
I want variety and divergence, not a flock of sheep in consensus about great literature. Best-sellers deserve some respect but should not distort results. Bomb-throwers should be offered books that appeal to terrorists, no censorship or central authority, effective privacy.
Negative feedback allows profiles containing books you hate to be applied in reverse, lowering rankings. Communication between Advisor library sites will increase the depth of the knowledge base, so clients with a particular, peculiar or exotic focus can find rare birds of a feather.
I believe that reading enriches existence. Experience weights our evaluation of reading as a rewarding activity. Better advice gives greater rewards, so more reading and a cumulative human benefit.
Al Hartley has 30 years experience with computers, mainly as a work-from-home software developer. He is a keen reader (particularly biographies) and has supported his local libraries in embracing the web. Al also has a continuing interest in psychology, studying at Macquarie Uni and numerous New Age programs (eg NLP).
visit dorrigo.com/advisor or email al@dorrigo.com