Sunday, February 20, 2011

Recommendations and Randomness: How I Pick Which Books to Read

There was a time when I went to a library or a bookshop and was immediately faced with a problem: how to pick which book to read?  Over time, I have changed how I solve this problem, which has changed what I read.

Maybe I should change again?

How Did I Use to Do It?
One technique would be to find the section that corresponded to an interesting-to-me topic and randomly browse through those books, trying to find one that looked worth reading.  I would look at the table of contents, skim the first page or two.  This was generally pretty successful  - and explains why I would read so many non-fiction books.

The other thing I would do is to see if I could find the latest book by an author I had previously read and liked.  Which would explain why I read so many mystery or science fiction books (genres which thrive on series and name-brand authors) and virtually no literary fiction.  Also, I like mysteries and science fiction.

Obviously, I was not alone in these techniques.  Bookshops and libraries continue to be organized by topic and type (although bookshops in America seem to be evolving into coffee shops and ebook device vendors, with paper books as something of an afterthought).  And the book publishing industry loves when a big name author releases the latest in a series.

However, over time, I decided to change things up a little bit.

Why Did I Feel I Needed to Change My Ways?
The revolution in self-publishing means that a lot more good reading is easily accessible online.  (Together with a lot more bad reading, too, of course - let's not forget Sturgeon's Law).  And it is not only accessibility which has changed.  The good stuff is a lot more discoverable, too.  For me, the advent of blogs (and later Twitter) has meant that I get directed to the best non fiction reading all the time.  More, in fact, than I can read.  (It is possible that I don't actually find out about the best non fiction online via blogs and tweets, but ignorance is bliss).

At the same time, the quality of non fiction books appeared to me to decline.  Too often, it seemed that the essential idea is captured in the title.  The rest of the book is then simply a rehearsing of that idea over and over, typically lacking nuance and depressingly often an over-simplified look at the world through a single lens.

My fiction technique held obvious flaws - how to learn about new authors or read something other than genre fiction?  In the immortal words of the Stranglers - Something Better Change.

I've Got a Little List
Really, the new-to-me technique I adopted is quite old and very obvious.  But here it is.  On the one hand, I extended the techniques for discovering the good stuff online to the world of books.  Once in a while, the blog posts and tweets would mention an interesting-sounding book.  So, I would add the author and title to an online list. (I use backpack).  But the real source of recommendations is The New York Times' Sunday Book Review (I read the dead trees version, but you may prefer it online).  Yes, I find that critics are the best source of information about books.  And, often for non fiction, you get enough of the idea from the review, so that you don't need to read the actual book.  (I try to supplement my source of critical information, including the Guardian, but the NYT is my main source).

Randomness and Routine
So, I now have a set of great recommendations for books and authors in an online list.  (At the time of writing, I have somewhere between one hundred and two hundred entries). But how do I actually select what to read?

Well, my normal routine is: go to the library, consult my list of books to read and randomly trawl through the list, consulting the library's catalogue until I find some that are available. I will also try to balance things out a little bit by trying to read some non-fiction, along with non-fiction. Plus I like to get at least one smallish, paperback book that I can read on the train. But random factors like who else happens to have checked out a book on my list, plus whether the library happens to have decided to stock it in the first place, heavily influence what I read in practice (rather than the theory of my recommendations list).

How is it Working Out?
I've been pretty happy with this set of techniques. I read way more fiction - and different types of fiction - than I used to. I tend not to read all that much non-fiction anymore. Although, having noticed this, I've resolved to do better with that. It means that I almost exclusively read books from the library (although I will read a book if you give or lend me it, too).

Areas where it doesn't work include books that are too obscure or controversial for my local library to stock. (Typically, this limits the non-fiction work I might read). And it also means that the idea of e-books (the Nook, the Kindle, etc.) doesn't make any sense to me. Half of my technique relies on the constraints of what is available at the library. I imagine that e-books have some constraints (maybe not everything is available as an e-book?) but it doesn't seem the same, somehow. Similarly, it means that I don't really visit first-hand bookshops anymore. Unless there's a specific book that I need to buy (generally for someone else) or it is some unique emergency situation (such as in an airport).

How do you pick your books?

Wednesday, February 2, 2011

Do OWL Classes Inherit Properties?

I'm slowly learning about using Semantic Web technologies. Sadly, I'm trying to do this in a rather ad hoc, as needed way, so my understanding is far from complete. I just ran across this interesting question: does an ontology subclass inherit the properties of its superclass?

Searching the web for opinions on this topic leads to a couple of contradictory views:
The answer to this question says: "Instances of subClasses do not inherit properties from instances of parent classes".
This states "In the example below, Penicillin is declared to be a sub class of both Antibiotic and USRegulatedMedication.It will therefore inherit the properties of those classes."

So, I turned to the W3C RDF Schema Recommendation, to see whether the definition would shed any light.
"The property rdfs:subClassOf is an instance of rdf:Property that is used to state that all the instances of one class are instances of another."

This reads to me that when C1 rdfs:subClassOf C2 then anything that is a C1 is also a C2.So, that seems to directly support the notion that C1 inherits all the properties of C2.

Certainly, if rdf:subClassOf doesn't mean that a subClass inherits the properties of the parent, then what does it mean? That it is unclear is a bit worrying, though.