Number of Books Toggle in Tag Mirror

I recently posted this same suggestion, but it got sidetracked with all sorts of other issues. Regardless of that I want, again to make the following suggestion.

Have a toggle in the Tag Mirror to display numbers of books for each tag, just as exists for the Tag Cloud.

If you go to the tag cloud for any book, the cloud there has a number toggle. In that circumstance those tags are in essence a tag mirror for that book.

The numbers toggle in the library Tag Mirror would be immensely useful to many members.

I have discussed this with Tim and because the Tag Mirror page is taking into account more than just the number of books, we feel it would be more confusing than useful for the number of books to be displayed right there on that page. Since the text sizes are calculated differently from that other tag cloud pages, we are not going to be putting the book counts on the Tag Mirror page. You can still get the counts by clicking on the tag and looking in the popup, but we don't want to confuse people by adding them to the cloud and making them think the text size is only related to that number like it is in other tag clouds.

>2 knerd.knitter: OK

>1 JoeB1934: I want to push back on the idea that the tag cloud for a work is "in essence a tag mirror for that book." A tag mirror is a cloud of the tags that other people have added for their books that are in the same works as the books in your catalog. It's a reflection of your catalog as seen by others. A tag cloud for a work is simply the count of tags applied to the books in that work in everyone's catalog. It's not a reflection of anything.

>2 knerd.knitter: I have now spent about a year working with the tag mirror for my whole library as well as individual collections. I have NEVER understood the text sizes as being useful to my interests.

What are my interests? It is simply to see the number of books in the collection that have been designated by other members as having that tag. Yes, I know that I can select any tag and the number is displayed along with the specific books.

I have a favorite set of 40 tags that I use to define my reading interests and the current technique of text size doesn't really give me much of a clue to the information that I want.

There might be some members who use the current tag mirror concept to great advantage and I don't want that to be changed. Why not have a numbers option for someone like me who simply wants to know how the books in a collection are arrayed?

The work tag cloud technique is fantastic and I can immediately eyeball for the data I want there. Is the programming too complicated on the tag mirror page?

>5 JoeB1934: Unfortunately, it is not possible to get the work counts on the tag mirror page before you click the lightbox. It would just take too long to get those numbers; it would make the page loading very slow. I've been looking at the code, and it only takes into account the 20 most common tags on the work for the purposes of generating the cloud. Then the lightbox shows the full list of them, which is why that number doesn't match any of the numbers I've attempted to add based on the data in the cloud itself.

>6 knerd.knitter: OK, I concede that my request is not doable. I have created an Excel model which allows me to do what I want for my whole library, or any collection. At my age I won't be doing this much longer.

However, I need to mention something that I have learned after working with these tag numbers for months.

This is the fact that the number of members who have read a book AND tagged it with any genre-like tags is shockingly low. I have seen situations where a book has a thousand members, but less than 100 have bothered to tag the book.

This means that prevalence percentages, which are usually (number of tags/ total readers) is not meaningful if only, say, 5 % of readers actually make tags.

In my estimation, this makes the tag mirror where the text size is based on prevalence of questionable value.

Do you have any idea about how many members are actually using the text sizes as part of their analysis?

>7 JoeB1934: I can't speak to your larger question, but I can tell you that since LT introduced the Genres feature I no longer bother using genre tags — it just seems redundant to me, unless I want to use a more specific sub-genre tag (i.e., I would use the tag "police procedural" but I would not also add the "mystery" tag because that's already covered by Genres). Of course, that doesn't mean that people shouldn't use genre tags — everyone's taxonomy is different and that's a good thing.

I haven't gone back and updated the several thousand books I had already used genre tags on before that feature was introduced, but as I do other cleanup in old records (updating covers, etc) that's one of the things I look at.

>8 rosalita: I use what I call tag-genres which are like your police procedural concept. in my analysis i use broader genres like mystery and literature just to provide information in cases where the reader stops at that larger level. This is how I have 40 different tag genres to describe my books. It is true that some of these are redundant to the standard genres provided by LT.

Each of us adds the tags that we find useful for our personal purposes. I see that you only use 40 different tags, but lots on each book. I have over 1,800 different tags, but usually only 1-4 on any single book.

>10 MarthaJeanne: I agree that each of us has developed a system for our own purposes. I don't have any what I call bookkeeping, or admin tags. I actually never create any tags for myself as all of mine for any book come from using tag mirroring to tell me how other members view my books.

The 40 tags possible are attributes for my books which I am most interested in. For example, one of my favorite tags is books-about-books. I then use the tag mirror on my books and find out which of them have such an attribute.

This is of great use for evaluating all the books in my TBR list to find which ones contain the books-about-books attribute, or any of the other 39 tags of most interest to me.

The reason I asked for this tag mirror feature was to assist me in assigning those tags for large numbers of books. I clearly can do it manually for any single book by going to the tag cloud for that book.

Tim doesn't want you to tag based on how others tag. He wants you to set tags yourself.

>12 MarthaJeanne: Yes, I know that from prior conversations, but I have learned an incredible amount about my reading patterns over the years and I don't see why Tim's preferences should stop me from something I find to be extremely useful.

Meanwhile, my observation is that a small proportion of readers actually do tagging. Also, I might say that if you look at a book in Goodreads, they produce a list of the numbers of all tags for a book and they automatically show for the book a list of about 10 or more primary tags for a book.

They show that list but provide an easy way for a reader to attach their personal tags to the book while knowing what other readers have done.

Let's say I have a TBR list of a hundred books I am thinking about reading this year. Wouldn't it be of great interest to know which of these books match a current sub-genre of interest, or a set of such genres that you prefer to read? The tag mirror of that collection does this in an admirable way.

Yes, I can go to the work tag cloud for each individual book and accomplish what I want, or I can go to tag mirror and click on my tag of interest and download all books with that tag. This is what I am doing now and exporting the tag count into an Excel spreadsheet which develops a sheet that shows which tags are associated with each book.

As I said earlier, I concede that LT will never make my process easier but I will continue to do what I need using tag mirror as currently configured.

No. But I suppose that is why I don't use tag mirror. I prefer to keep my reading very diverse.

One final statement that I can make is that after all is said and done when I finish a book the tags obtained by using the mirror is always spot-on in my judgement. I learned about magical realism after the mirror results told me a significant fraction of the books I like contain that attribute.

Another important point for me is that I use the tag mirror to find out if a book has any of my favorite tags. If it has a mention of a specific tag I simply acknowledge that as a connection. I am not looking for the tags with the highest count. So, I am using only my tags when I say a book does reference that tag. In my world all tags have equal value to me and a given tag can be associated with a book at a low level as long as it is one of my favorites.

Thanks for a nice discussion of how to use tags in different ways. I have genre tags i.e. "mystery" and some time related, i.e. "1945" and some publisher series where the name of the series makes it hard to search for just those books. But looking at other LT'ers tags gave me some new one like "Epistolary novel" that I hadn't considered before.

I found a way of getting (a snapshot) of the tags other LT'ers use on my books, so something similar to what >1 JoeB1934: asked for, I think.

Here's the result in case someone gets inspired by it:
https://wiki.librarything.com/index.php/User:Bnielsen/TagShort1

>16 bnielsen: I should have known that you would come up with this suggestion as you have had excellent solutions in the past!

You are correct that the output you generated for your books seems to solve my request as well. The only question is how I, a non-wiki person can do this also. My library is around 1700 books so there isn't a volume constraint for me.

I am curious as to your results for any book you want to evaluate. Does your list for that book match the tag cloud for that book work page?

Thanks for your contribution.

I've taken a look at my script and it doesn't really know anything about my books except book_id and work_id.
The key is to use the ajax interface: https://www.librarything.com/work/$workid/work/$bookid/ajaxinc_showbooktags.php&...

all=0 gives you the short version. all=1 gives you all tags - i.e. probably too much information :-)

So I took a look at your catalogue
https://www.librarything.com/catalog/JoeB1934
in a style that contains book_id and work_id and copy/pasted the 9 pages into a text file and extracted the book_id and work_id columns, giving a list of 1713 entries.
I then ran my script on that list and put the output into a new wiki page.
https://wiki.librarything.com/index.php/User:Bnielsen/TagShortJoe

The output should match the tag cloud for that book work page. (If you go to that page, you can toggle between showing all or some of the tags). If you choose "all" you'll get some very large tag clouds for some of the books.

So to sum it up: I hope you can use the data on TagShortJoe for something useful and maybe share some of it here?
The format of the list is SPACE book_id SPACE::SPACE tag1(count1) SPACE::SPACE tag2(count2)
I hope you can manipulate this in Excel to get whatever you want :-)

My script is still running, so I'll upload a part of the list and then the complete list when the script is finished. ETA: the script finished and I've updated the list, so all 1713 books are included)
Let me know what you think of it.

>18 bnielsen: I just woke up at normal 4AM MST and am reading your post. Thank you so very much! You have done a lot of work and I really appreciate it.

Yesterday I started a review of my Excel model and I think there is a reasonable way for me to incorporate a file structured like yours. It will probably take me a couple of days to determine how best to do so.

First, I will eyeball the file you created. I have converted your file by copy & paste into an Excel file and scanning the output I can recognize records even though I don't know the book_id for any of them. I can do a LT excel export and get the book_id,title and author so I will be doing that first to convert your file almost as if it had been created as part of the normal LT export.

I have yard work and a doctor visit about my sciatica today so might not have more to report today. but will update on progress.

Thanks again!

Oh! I also forgot workers are coming to install solar panels on my daughter's house, who I live with.

>19 JoeB1934: Nice. Thanks for taking the time to report on the progress so far. Eyeballing is always a good thing, so that's why I used " :: " as separator. Eyeballing made me look at one of the books that is tagged with
"no sex scenes" as well as "characters that have sex" :-) By the same user it seems.

https://www.librarything.com/work/23922684/244369007

(Hmm, looking at that I notice that "Numbers" is a toggle, but "Show all" isn't.)

Thanks for making me look into this. It took me a bit of time to realize how simple Ajax is :-)

If you want to play around with the full tag mirror for all 1713 books just say the word. The results will be too large to fit onto a wikipage, but I can easily just send you an excel file.
I took a look at the full tag mirror for some of the books and one of them came to 80 Kb, i.e. a lot more than all the other information in the LT export file on that book. I'm not sure I could use it for anything, so I was fine with the short version.

One last finding: Looking at the work page for a given book, you'll see your own tags on the top of the page. But even if you look at the full tag mirror, some of your tags might not be there. That's because the tag mirror shows you the tags after combination.

Ah, some of the tag counts contain commas, which my script didn't expect, so it gives us stuff like
:: fiction (2,154) ::
i.e. some extra spaces and the unwanted comma in the count. I want
:: fiction(2154) ::
so I'll fix that soonish.

>21 bnielsen: Not to worry as my current approach just counts on the main tag name. Have had time to do a bit of work and am pursuing a track which I hope can come to fruition tomorrow.

I converted your file into an excel file with book_id as first column and every :: created an excellent break-out of separate columns for each tag name.

So, I can search column 1 for a specific book_id and search horizontally for a specific tag name. I repeat this for each of my favorite tag names and I end up with just the tag names I focus on for each book.

>22 JoeB1934: Nice. That's what I was hoping for. I've fixed the comma weirdness and updated the wiki page.

https://www.librarything.com/catalog/JoeB1934

Let me know if you see other weird stuff :-) The tags are 99% provided by others so I tried to construct my script to be rather robust, but you never know!

I wrote a script to analyse my own tags. I.e. if all books I've tagged with Geometry are also tagged with Mathematics, the script will report that Geometry seems to indicate Mathematics.
I'll try running it on the TagShort data sometime.

>23 bnielsen: Singleton tags, i.e. tags that only occur once in the tag mirror dataset, i.e. https://www.librarything.com/catalog/JoeB1934 gave a lot of uninteresting stuff, so I told my script to skip them. Now I get stuff like this:

Jo Nesbo implies Harry Hole
Jo Nesbo implies Kindle
Jo Nesbo implies Krim
Jo Nesbo implies Norway
Jo Nesbo implies Norwegian
Jo Nesbo implies Norwegian literature
Jo Nesbo implies Oslo
Jo Nesbo implies Scandinavia
Jo Nesbo implies Scandinavian
Jo Nesbo implies crime
Jo Nesbo implies crime fiction
Jo Nesbo implies detective
Jo Nesbo implies ebook
Jo Nesbo implies fiction
Jo Nesbo implies murder
Jo Nesbo implies mystery
Jo Nesbo implies novel
Jo Nesbo implies police
Jo Nesbo implies police procedural
Jo Nesbo implies read
Jo Nesbo implies series
Jo Nesbo implies suspense
Jo Nesbo implies thriller
Jo Nesbo implies to-read
Jo Nesbo implies translation

so if a book (selected among the 1700 owned by JoeB1934) is tagged "Jo Nesbo" by someone it is also tagged translation by someone. And it seems that all of the Jo Nesbo books are police procedurals.

>24 bnielsen: That is very interesting! I will have to consider the meaning of that. One of my variations is to look at all tags by an author. For example, my most favorite author is Kate Atkinson and I often find her tag set among all of her books in my catalog but it would be terrific to find all her tags for all books across the LT world.

In many ways an author tag set is more important because readers aren't very reliable and inconsistent so the more readers involved in tagging an author, i believe would smooth out the tags. I happened to have read all books by certain favorite authors but not by the majority of my authors which number around 700.

I have incorporated the bnielsen tag file into a test excel model of my library. Early indications are that, on average my books have about 6 tags per book with the new method. The average was about 9 tags in my original method.

I have always used any tag even if the tag count was quite low. This result might be a resultant of using a 'short' count?