If you're not already... be sure to get subscribed today!
Tip of the week
Building concepts not keywords
One thing that search engines do a lot of the last while is look for categorizations. This can be for relevance, or simply to speed up processing. They can use it for personalization, query analysis, semantic relations and more. I’d go as far as looking towards page segmentation concepts to understand exactly how granular this may get some day.
What this means to the avid optimizer is that you really want to build strong themes in your over-all programs. When you have a variety of separate concepts within a site, be sure to define them through structure and targeting core and related terms (keywords). When building logic trees within the site, think in terms of themes and concepts.
- Where we once thought (and still do) in terms of building terms outwards; ‘SEO’ ‘SEO Services’ ‘affordable SEO services’
- We want to add a layer of; ‘SEO’ ‘LinkBuilding’ ‘Keyword research’
As you can see they are definitively different targeting approaches. That is because we’re not using modifiers in as much as semantic theme development. This can be used structurally (urls, page meta data) it can be used on-page as well as in the link building program. Creating strong themes/concepts is an important aspect of modern SEO….
Talk of the town
|Link tools - One thing is for sure, I am not a huge fan of tools Sure, I use them like anyone else, I am simply suspect of the data. In the end it is the analyst not the tool that matters. They need to ask the right questions to find the right data. To that end I enjoyed Michael’s; The Problem With Backlink Checkers And Backlink Analyzers - |
Kumo and search evolution lesson – as some of you may have heard, Microsoft is yet again looking for ways to increase their search share; this time with a rebranding, code-named – Kumo. Here’s the chatter from their blog;
I really liked Danny’s write up of Kumo as well, as it is a nice walk down history lane.
|Michael does page segmentation – over on the SEO Theory blog (once more), Michael outlined an interesting relevance test that bears a striking resemblance to page segmentation concepts. Not only is it some interesting ideas, it shows why search engines like block level analysis for finding relevance points in a page. |
|Google testing new site-links? Barry was reporting that some new site-links were seen out in the wild (horizontal, not vertical). The new method makes them less obvious/intrusive and I had to wonder if could click bias play into it? Are the standard site links incurring to high a CTR thus skewing the rest of the results? Quite possibly… |
|Matt Cutts; link spam assassin – was talking (once more) about pay-per-post type link manipulation techniques. While the tin-foil-hatters were going on once more about Google’s morality police, fact remains it’s tricky business. |
Pay for Performance – Stoney had a good article on something I would NEVER DO… which is a paid for performance model.. my fav line is “Every time I've gotten into a pay for performance model I got burned. Each time I got burned it was due to having a lack of control over the site.” – It is definitely worth a read.
Search Geek Central
DocJax is a search engine for documents, which allow you to search documents and e-book from everywhere, preview them and even download them for free.
Matt Cutts did some grab-bag questions and even touched on the ‘Brandy’ update (or 'Vince Small Change'). As expected he speaks of changes in some query spaces, but nothing ‘brand’ specific was done. You can find the video and complete transcription of Matt's comments on the Trail;
Behavioral targeting control - while not entirely search related, the evolution of behavioral targeting is the subject of this video that speaks to where it may be headed ultimately.
Google + Twitter – here’s an interesting Grease monkey script to add top 5 Tweets for a given search term.
Htaccess cheat sheet - Search Engine Journal had some interesting cheat sheets, including this one for htaccess which is always handy
10 Must track Google Analytics goals - this one is just like the title says. Analytics are more important than ever these days, learn to use these weapons well.
Seriously Obsessed Search Geek Alerts
Search Patents - March is starting off reasonably well for the patent geeks and we had a few interesting ones this week;
Google – Interface for a universal search - this one, from August 2008, is good if you have an interest in Universal Search results including how they determine popularity of verticals for a given query.
Google - Word decompounder – filed Sept 2007 - Last week we had a natural language processing patent on geo-targeting and another on dealing with languages such as Chinese. This week there is a Google patent dealing with languages that utilize compound words (such as German). It deals with taking them apart to better understand the query in question.
Google - Search filtering – filed August 2007 – this one deals with (somehow) filtering out undesirable content (such as porn to juveniles) and having a service that deals with verification. I didn’t read it all, but what I did was strange enough to at least mention (lol)
Yahoo - Automatic extraction of semantics from text information – originally filed in August of ’07, this patent deals with image search (can you say Flickr?) and assigning semantic concepts from tag and other meta-data. I like this one, there haven’t been a lot of image search related goodies of late.
Yahoo - Automated most popular media asset creation – once again from August of 07, this patent looks at creating lists of ‘most popular’ media assets such as video and images.
Yahoo – String processing and searching using a compressed permuterm index. - Interesting one, also filed in August ’07, that deals with optimizing string dictionaries (tolerant retrieval issues). This one of for the geeky types that are interested in how search engines deal with ever growing databases of words from which to understand and process queries better.
Microsoft – Presenting result items based upon user behaviour – filed in August ’07 this patent I pretty much just as it sounds. Of interest with this personalized search approach is the highlighting of the personalization’s.
Our resident information retriever (and SOSG name-sake) shall be helping us keep an eye trained for interesting IR goodies. Here’s this week’s offerings –
Search engines and long tail queries; The lady CJ herself had an interesting look at some research using Microsoft data that highlights shortcomings in modern search engines. Essentially, due to a lack of NLP abilities, the longer the query, the less relevant the results are.
n-gram words an language Ordering model with – is worth including if for the title alone (hint, word ordering is mucked). I’ve said it a few times, SEOs should be watching the natural language processing world – it’s the future of search.-
Generalization Bounds for Ranking Algorithms via Algorithmic Stability – an interesting optimization approach for machine learning. (PDF)
/end SOSG session
I am definitely interested in any feedback you may have or ideas for what you’re interested in – Please feel free to get in touch
( ad if you reading the archive... be sure to get suscribed today! )
|< Prev||Next >|