MobileRead Forums - View Single Post

Tex2002ans · 11-08-2014, 01:37 AM

Quote:

Originally Posted by Hitch

Creating competent indexes is not simple. It's not easy. It's not an "overlooked opportunity." It's a lot--a LOT--of work to do well; it's easy to do badly [...]

Oh yeah, if we are talking building an Index from scratch, it is absolutely BRUTAL to create an Index. If you already have the print Index though, AND you have your EPUBs marked up with page ids, it wouldn't be TOO bad.

This is the way I have seen it done (this is a real sample from an actual ebook with a linked Index):

Quote:

<p>[...] The study of history makes a man wise <a id="page31"></a>and judicious. But it does not by itself provide any knowledge and skill which could be utilized for handling concrete tasks.</p>

[...]

<p>[...] Where there is something to explain, the human mind has never been at a loss to <a id="page32"></a>invent ad hoc some imaginary theories, lacking any logical justification.</p>

As you can see in RED, throughout the book is a <a id="page###"> that matches the page number in the physical book.

Then, in the Index, you just map every single page number to its equivalent link. So you change this:

Quote:

Dialectical materialism, 79–84
Deflation, 419–421, 428, 564–567, 779 see also, Cycle theory; Depression; Money

into this:

Spoiler:

As you can see, the code can get real hairy real fast. And there is currently no automated tool that I know of, that can do this (you COULD probably come up with some regex to help speed up the process).

Side Note: Perhaps it would be fastest to merge the entire book into one mother XHTML file, create some explicit links to convert those page numbers into "page###" links, and THEN split the book into its respective chapters. That would let Sigil deal with the bulk of the spaghetti link renaming.

I don't see a linked Index using this method, as being any worse than reading the physical version of the book.

Physical: You look up page number, you turn the ancient stone tablet over to that page, you scan page looking for where this word/person was mentioned.
Digital: You click and jump to the "page", you scan the page looking for where this word/person was mentioned, AND/OR you just search for the word yourself.

Quote:

Originally Posted by Julanna

A solution to that could have been to have a temporary highlight on the target in an ebook when the link it activated, and that's what I've realised, on this thread, isn't possible with current html/css and ebook conversions.

I don't know, I think this would be better handled at the READER level itself... Perhaps something like Chrome's Ctrl+F search, where the scrollbar turns into yellow highlights showing off the portions of the page where the search term occurs:

http://cdn.raizlabs.com/gregshead/wp..._scrollbar.png

Perhaps one of the non-mainstream devices/readers like Marvin, or the multitude of Android EPUB readers could implement something like this? Could be a nice selling point over the competition! :P

I believe AZARDI just implemented entire library search. Where you are able to search the text of every single ebook added to the program... I would believe this to be infinitely more helpful than a stinkin' Index. Sort of like a Google (Searching) compared to that ancient Dewey Decimal System (Index).

Quote:

Originally Posted by Julanna

And, actually, having multiple targets is easily displayed in the index using a, b, c etc. Looks good that way. The code part of the targets in mine also have letters that match. I use find so I don't miss an instance of a name and paste code down the page, just changing the letter difference as I go. I don't have trouble knowing where I'm up to and then I can create the correct number of links in the index just changing the letter on each.

Oh boy... I don't know what sorts of Indexes you are looking at, but there is just no possible way that would fly in books that I have seen. Perhaps that might work in the simplest of books, but it could get quickly WAY out of hand.

For example, this text in the physical book:

Quote:

Dialectical materialism, 79–84
Deflation, 419–421, 428, 564–567, 779 see also, Cycle theory; Depression; Money

Converted to this with the "page number" equivalent:

Spoiler:

Compare this to what Sigil's Index Tool created:

Spoiler:

Now I have to go through, and check all 14 usages of "Deflation" and all 2 usages of "dialectical materialism". Now I have to compare this to the physical book's Index.

Now I have to double-check every link to make sure that where it points to is actually RELEVANT, AND falls in the text that the physical Index referenced (for example, I have to find that Sigil's new "dialectical materialism" links (1, 2, 3, 4,...) actually FALL in between the text on pages 79-84. Now I have to make sure that all 14 entries of "deflation" falls into the page numbers referenced in the Physical book's Index.

Each one of these comparisons takes AT LEAST 20 seconds (probably much longer) to double-check and read the sentence/paragraph to see if that spot is relevant. Let's say I can get 3 done in a minute (being generous), with an Index of hundreds/thousands of links, this adds up to HOURS and HOURS of work.

As you can see, the work that actually has to be done, to create a PROPER linked Index, goes exponentially THROUGH THE ROOF. As Hitch said, it is very easy to create a crappy Index, but to DO IT RIGHT, it is absolutely brutal. And what is the entire point of wasting all the manpower on that crap, when you have a full SEARCH at your fingertips.

Edit: Actually, the way I did Sigil above only found the capital "D"eflation, if I added in the lowercase variant, the amount of links in Sigil's Index jumped from 14 to 40. So that would be, on the extremely low end, at least 13 minutes of work just for one word.

Side Note: This isn't even covering all the other work/manpower that has to go into creating a great Index (it really is an art, and I don't feel like writing another tome

). For example, you have to handle multiple variations of the same word. In the physical Index, it might just say, "Speculating", but you have to cover the words "Speculate," "Speculates", "Speculation", "Speculative", .... Now that "13 minutes for one word" estimate? Yeah, that goes much higher.

How do you handle when the name/word is referenced multiple times within a short span? Do you just point to every single line/paragraph the word is mentioned? What if "Deflation" was mentioned in 10 paragraphs between pages 419-421? Are you going to have 10 links that jump you to "the same" general location?

Quote:

Originally Posted by Julanna

Ok, and saying again, I know it's the creation process. The creation process is what needs to change.

[...]

At this stage I don't care what publishers will pay for.

Most of the publishers that I have seen, are now just completely chopping out Indexes in their ebook versions, because they are obsolete and make zero sense in an digital book. (Similar with this preposterous idea of "page numbers", they make about zero sense in ebooks + websites.)

These are relics of the physical age, and must go the way of the dodo!

Quote:

Originally Posted by Julanna

Search is not as good as an index, an index is a structured finding tool that facilitates learning.

How do people learn things on the internet? There is no Index there to help guide you! You just SEARCH for all the info you need.

Anyway, my personal philosophy is, I do not waste my time on creating those insane page links + linked Indexes (it is just not worth the time/effort). Although I personally DO leave the actual text of the Index in the book. Why throw it away if someone already painstakingly categorized the book?

Leaving the text of the Index in the EPUB, allows a person like you to still take a look (as you would in the physical book), and then you can easily SEARCH for those words yourself if you really wanted to. But all this linking back/forth/every which way? Nah, no thanks.

Quote:

Originally Posted by Julanna

[...] Should have been written up with a more helpful tone.

Hard to convey emotions through text, typically you interpret the words with whatever your current mood is (if you are angry, you might feel that the posts on MobileRead are especially infuriating/mean)... maybe that is why I sprinkle smileys all over (ok ok, I lied, I use them excessively

) ... I don't think you could interpret a smiley face wrong!

So what have we learned today. In order to teach everyone about how horribly labor-intensive Indexes are, we channel "happy thoughts"!