Forum Moderators: open
From The Anatomy of a Search Engine [www-db.stanford.edu]:
Aside from PageRank and the use of anchor text, Google has several other features... Google keeps track of some visual presentation details such as font size of words. Words in a larger or bolder font are weighted higher than other words.
Then there's this on the Google AdSense Overview page:
[google.com...]
...an algorithm that includes such factors as keyword analysis, word frequency, font size, and the overall link structure of the web, we know what a page is about...
I'm not advocating anything. It seems that this is a little known thingy out there that not many people are aware of, so I thought I'd toss it out there.
Or they now parse CSS because otherwise their is no way to know how big the font really is.
What do you think?
Leo
We use font size relative to the rest of the document because when searching, you do not want to rank otherwise identical documents differently just because one of the documents is in a larger font.
By the way, there were two significant changes in the weight of the font size as well as other on-page factors for Google this year. During Domininc and during Florida the relation between bold, italic, headings, font size were changed.
During Domininc and during Florida the relation between bold, italic, headings, font size were changed.
could you expand?
Leo
As long as Google doesn't change it's algorithm, the order of the pages (for a specific keyword) should be unchanged even if PR is changing or the weight of off-page factors is changed. However, I saw changes during Dominic and Florida.
How can you have two exact same pages to verify this? There should always be some kind of difference even if small.
With the Florida Update word stemming has been introduced therefore a variant of your word may be tempering your test.
By font size I assume they meant <h1>'s, <h2>'s, <hx>'s.
how to "weight" font/text without
Or they now parse CSS because otherwise their is no way to know how big the font really is.
Which would've be my question too and which I feel hasn't been addressed/answered here yet ;0
What about <strong>, <em> as well as the more widely used <b>, <i>, <big>, <font size="6">, <Hx>
AFAIK CSS isn't being parsed yet, so how would they know the "font size of words" without taking their cue from any of the the above "tags" whether or not they've then been styled with CSS to look different than expected..
<span class="x">font weight</span>, where "x" is externally specified as font-weight: bold;, I presume won't work until CSS is parsed, but by then I imagine that this "game" will well and truly have been played out ...
Therefore, back to basics, I think that only the tags that "say" what they do will count whether or not, or until(?) CSS is parsed.
To me it makes more sense that if they do want to rate the page on the "font-size or weight" then it might make sense for them to "deprecate" (or in their case, give less weighting to) the <b>, <i>, <big>, <font> elements as XHTML will probably do.. that is they are keeping up with the times, albeit for different reasons ;)
I've read that <b> was used in their algo before, it maybe still is, but that <strong> wasn't recognized, well maybe it is now?
(X)HTML is moving on as are the algo tweaks so I think it's not so clear cut as it was in that first article (1996 was a long time ago in web years!).
There are many ways of "bolding" text and *ALL* of them are open to abuse so my conclusion is that whatever way a web page author chooses: external CSS, <b>, <strong> or <Hx>, is becoming an outdated SEO tactic purely for the reason that the abuse of these are becoming so widely known by SE algos (thanks in part to WW;)).
I suggest they (SE's) will be weighting the page with some sort of quantity of "large/bold" on page text versus normal font-size/weight density algo, therefore wrapping the entire page, or indeed an entire paragraph in <h1> or <strong> tags will reduce its effectiveness entirely!
and so using CSS or <font> tags is not going to matter.
I can't see that they can penalise some sites because they keep up with standards and don't use <font size="6"> anymore, or conversely I also can't see how they can penalise sites that use <strong> or <em> to correctly (accessibility wise) add emphasis... the same goes for all the above mentioned tags except perhaps <hx> which hasn't been replaced yet ;)
perhaps it's a case of the toy has been abused too much , now they're taking it away
As for parsing CSS, if they get the above bit right first, they'll no longer have to worry about parsing CSS to help overcome any "abuse" of font size/weight problems ;)
Suzy
How can you have two exact same pages to verify this? There should always be some kind of difference even if small.
In most of the cases a different page number should be enough. Changing this number shouldn't affect any parameter of the ranking algorithm such as keyword density etc.
With the Florida Update word stemming has been introduced therefore a variant of your word may be tempering your test.
Stemming doesn't influence my results, because
- the pages aren't in English
- I'm looking only for order of my pages (not the relation to others)
- of the search words and text used
CSS may not be parsed and reviewed, but a SE could look at content and read <font size=+1> which would make that text larger than the rest of the page it is crawling through.
A parser is defined as an algorithm that determines the syntactic structure of a piece of information (e.g., sentences, words). A web crawler is an algorithm that finds and extracts information from specific web document elements (links, anchors, titles, etc) while moving across the web.
Based on my research into the subject of how spiders index html/xhtml, I'm assuming that font-size has no relevance. As pointed out above, there are too many factors involved for a crawler to determine font-size into the equation based on html tags.
The crawler is programmed to index content between start and ending tags. I'm not too certain that it would be looking at any styling code within in the tag. It's just looking for (<) and (>) along with any other attributes it has been programmed to index, i.e. alt="".
My understanding so far tells me that crawlers are looking at the structure of the document and can assign relevancy based on the types of elements being used to structure the document. I would think it sees <h1> thru <h6> as they were intended, from largest to smallest with <h1> being the largest and <h6> being the smallest.
MartiniBuster, it would be an interesting experiment to see how two pages; one with the <font-size="2"> tag as opposed to one using styling through CSS.
P.S. Google has been spotted indexing CSS files. There were some topics on this earlier in the year.
it would be an interesting experiment to see how two pages; one with the <font-size="2"> tag as opposed to one using styling through CSS.
My first thought is, "Geez, that would sure look ugly."
But I'm curious and bored the way a kid is when he pulls the wings off a fly.
