WalMart Blasts Off with Polaris: Destination Semantic Search
August 31, 2012
I just read “WalMart Rolls out Semantic Search Engine, Sees Business Boost.” The semantic technology “not only helps users find items they want on its Web site, but also delivers results based on their interests and likely intent, the company said Thursday [August 30, 2012].”
Let’s begin with an essay question. What is semantic search? You have 15 minutes, describe how “semantic search works,” name three vendors with profitable semantic search businesses, and identify the strengths and weaknesses of semantic search, including costs and computational requirements.”
I assume you skipped the essay question. I know that Martin White and I struggled to craft a brief, reasonably accurate definition of “semantic search” in our Successful Enterprise Search Management, published by Galatea an eon ago. I think the book is still available at www.galatea.co.uk.
I had to define “semantic search” in my new monograph for Sue Feldman at IDC, and I think I recycled something from Wikipedia. The definition is okay, but I am not comfortable with that definition or any of the definitions I have written over the past few years.
The reason is that “semantic search”, like big data or analytics, is pretty much meaningless. Depending on whom you consult, the speaker trotting out the term, or the company touting its “semantic search” system—there are too many angles on the topic.
Find someone else to grade your exam, pilgrim. Or, better yet, just give yourself an “A” and move on. Easier and in today’s search environment, good enough.
I thought about semantic search when I worked through the flurry of “real” journalists’ and “real” pundits’ writings about Wal-Mart’s semantic search system. You know WalMart, the outfit that made Fairbanks Second Street into a ghost town overnight.
“Wal-Mart’s Homegrown Search Engine Already Paying Dividends” provides the insight which makes my beta blockers work overtime. The article reveals that WalMart has written its own search engine “from scratch.” In today’s world with the open source options readily available, I wonder what “from scratch” means. The story reveals that Endeca is being replaced with Polaris. Endeca, in case you are curious, is one of those late 1990 search engines which has been gobbled by a giant company. In an effort to pump up Endeca’s revenues and pay the estimated $1.1 billion acquisition price, some folks may be worrying about the total cost of ownership of Endeca. I am okay with Endeca but Oracle may not be happy with the pace of revenue growth. ( had heard that Wal-Mart had a brush with Google’s search appliance, but I don’t know if that technology delivered what Sam’s folks needed.)
WalMart offers fewer items, so, the article reports, WalMart had to figure out how to make search work better. The key point in the write up is in my opinion:
The new search engine technology has rolled out to both the Web site and mobile site in the U.S. Wal-Mart is now planning to roll out internationally to Brazil and other countries. @WalmartLabs was created in part by the $300 million acquisition of Kosmix, a data company based in Mountain View, Calif.
Kosmix had a search system, so I am curious about how “new” the technology is which WalMart is using. Kosmix was an interesting system and the company had, at one time, some interactions with Google. (For more information about Kosmix, see “Kosmix: YAGK (Yet Another Google Killer)” and “Kosmix and Its Positioning”.
Another outfit covering search is the estimable Technology Review. “Wal-Mart Dives Into Search Technology” is representative of similar stories in many blogs and trade news coverage. The main idea is that WalMart is aware that WalMart people are going online to buy the treasures from Wuhan which are available in the physical stores. The article makes this point, which strikes me as something quite a few people overlook:
All in all, focusing on search seems like a good move for Wal-Mart, but to fully see the benefits, the company will have to bring shoppers to its site in the first place, rather than competitors’ such as Amazon.
Do I have an opinion about Polaris? Well, owning a search engine sure looks more economical in the long run. Is it? Wal-Mart will know soon enough. And some of the people who are likely to use Wal-Mart’s Web site may just compare prices against Amazon’s and make a decision based on price. WalMart people without computer shopping in their DNA may just grab a cart and cruise a store. In Fairbanks, small towns run a bus to WalMart for shopping. No online connectivity in some places where WalMart has a store. We’re watching.
Stephen E Arnold, August 31, 2012
Sponsored by Augmentext
Clearing Up or Adding to the SharePoint 2013 Rumor Mill
August 31, 2012
As rumors continue to fly regarding SharePoint 2013 it is hard to say at this point whether they are being dispelled or expounded. ShareMuch is adding their two cents with a Screencast video specifically targeted at SharePoint search and FAST search. Watch the full video, “Screencast: Clearing Some Rumors Around Sharepoint 2013 Search and FAST Features.”
The author gives us some points to be on the lookout for as we watch:
Here are some highlights of the video:
1. Search service is still a service application
2. Configuration options in Central Admin UI are pretty much the same as in SharePoint 2010 search
3. Content sources can be defined on the site and tied to a rendering template
4. Query Builder is a neat feature to create query transformations and preview them before publishing
5. Importing and exporting search configurations is great
SharePoint 2013 is exciting. The enterprise search community can hardly wait to see what’s around the corner. But for some smaller enterprises, a smart idea might be a third party solution. Fabasoft Mindbreeze Enterprise can work as a standalone solution or compliment an existing SharePoint infrastructure. Updates are released quarterly and require no maintenance or hassle on the part of the organization.
Philip West, August 31, 2012
Sponsored by ArnoldIT.com, developer of Augmentext.
SEO Pro Emphasizes Content Quality
August 31, 2012
Desperate SEOs are constantly looking for a new angle. In Search Engine Journal’s “Content Confusion: Why Webmasters Fail—Time After Time,” Gregory Smith tries to convince his readers that it should be all about quality content instead of the latest ploy to game an algorithm. He begins:
“I’ve discovered many Webmasters who have lost a lot due to Google’s Panda and Penguin updates. Many have lost their entire income virtually overnight. Can you imagine how badly losing your income would affect your life in a very intolerable way? Instead of learning from this lesson, more and more people are steadily heading in the same direction, still completely confused.”
Yes. Smith spends some time discussing ranking tricks like guest posts, links of dubious quality, and infographics and the very good reasons such tactics no longer work like they used to. Though he insists that search engine optimization is still “one of the most valuable investments that any business owner” can make, he implores his SEO colleagues and their clients to focus on providing quality content. If they do, the rankings will take care of themselves, he says.
Well, to a certain extent. We wouldn’t want all those SEOs out of a job just because their field is becoming increasingly obsolete, right?
Cynthia Murrell, August 31, 2012
Sponsored by ArnoldIT.com, developer of Augmentext
Twitter Politics
August 31, 2012
Oh, goody, more predictive silliness. TechNewsWorld informs us, “Twindex Tracks Pols’ Twitter Temperatures.” Clever name, though it does make me think more about window cleaning than about politics. That’s ok; window cleaning is the more engaging subject.
The full name of the metric is the Twitter Political Index, and it tracks tweeters’ daily thoughts about the two presidential candidates. Twitter created the index with the help of Topsy Labs and pollsters at the Mellman Group and North Star Opinion Research. The polling firms helped validate and tune the algorithms. It is Topsy’s job to track tweets for certain terms and compare sentiment on each candidate. So far, the incumbent seems to be well ahead in the Twittersphere.
But how far can we trust the Twindex? Probably not very far. Writer Richard Adhikari observes:
“The Pew Research Center has found that only 15 percent of adults online use Twitter. On a typical day, that figure is only 8 percent. . . .
“Overall, nearly 30 percent of young adults use Twitter, up from 18 percent the previous year. One in five people aged 18 to 24 uses Twitter on a typical day.
“Further, 11 percent of adults aged 25 to 34 use Twitter on a typical day.
“African-Americans are also heavy Twitter users, with 28 percent of them using Twitter overall and 13 percent doing so on a typical day.
“Urban and suburban residents are also significantly more likely to use Twitter than those in rural areas, Pew found.”
So, yeah, statistically Democrats are likely to fare better among Twitter users than Republicans. This index is about as valuable as any political echo chamber—for entertainment only. Personally, I’d rather be washing windows.
Cynthia Murrell, August 31, 2012
Sponsored by ArnoldIT.com, developer of Augmentext
Wikipedia Studies Itself
August 31, 2012
We think studies about oneself are fascinating. TechEYE.net shares our enthusiasm in “Wikipedia is Accurate Says, er, Wikipedia Study.” Last autumn the Wikimedia Foundation tapped Epic, an “e-learning” company, and researchers at Oxford University to perform an assessment of Wikipedia’s accuracy. The results of the reflectively funded study? Wikipedia was found to be more accurate than Encyclopaedia Britannica. What an upset! Writer Nick Ferrell notes:
“For the record, if you wrote a page on Wikipedia about yourself, you would find that one of its teams of editors had deleted it for being advertising. However when Wikipedia commissions a study into itself and reports that it is wonderful, this is apparently ok.”
Apparently. Incidentally, a 2005 external peer review showed an average of four mistakes per article, as compared to Britannica’s three. The free encyclopedia has improved markedly, it seems. The new report also found Wikipedia articles tend to be more up-to-date. No surprise there; I’ll give them that one, at least.
Ferrell observes:
“What makes us smell a rat is that the report said that there were little differences between the two on style and overall quality score. We were not aware that the Encyclopaedia Britannica articles were penned by a person with a crayon, like some of the Wikipedia articles appear to have been. Nor does the Encyclopaedia Britannica employ people with faked doctorates.”
Good point. I think I’ll wait on an objective study before I draw any conclusions.
Cynthia Murrell, August 31, 2012
Sponsored by ArnoldIT.com, developer of Augmentext
EntropySoft Offers SharePoint Users Aid
August 31, 2012
SharePoint is a big name in the semantics tech world, but unfortunately it is not a one-stop shop. For businesses who are trying to utilize SharePoint’s features, the not so savvy user might need to access the program information, making things complicated to say the least. The article, Search and Act in SharePoint explains how using extra tools like EntropySoft’s FAST Search makes for a better user experience:
“EntropySoft’s FAST Search for SharePoint and FAST ESP Connectors allow documents from external content repositories to be included within their indexes. This means users can securely search and find documents from these external ECM systems directly from the native search bar in SharePoint…Users can access document versions and metadata, and with the appropriate permissions, they can also edit meta data or delete versions.”
The solutions to some of SharePoints most pesky user issues definitely allows for a more enjoyable experience using the program. In an ideal world our BI technologies would be advanced enough that there wouldn’t be a need for extra tools like EntrophySoft’s Fast Search or their other child EntropySoft Content Hub SharePoint Edition, a tool that allows an easy to use single point of access – but, we will take what we can get.
Edie Marie, August 31, 2012
Sponsored by ArnoldIT.com, developer of Augmentext
Google Updates the Portal from 1996: Info on Multiple Devices
August 30, 2012
The portal never really died. AOL and Yahoo have kept the 1990s “next big thing” despite the financial penalties the approach has imposed on stakeholders. There are other portals which are newer versions of the device which slices, dices, and chops. Examples I have looked at include:
- NewsIsFree, which delivers headlines, alerts, and allows me to find “sources”
- WebMD, which is a consumer “treat thyself” and aggregation information portal
- AutoTrader, which provides a vehicle research, loan, and purchasing portal.
Google when it rolled out 13 years ago took advantage of search systems’ desire to go “beyond search.” The reasons were easy to identify. Over the years, I have enumerated them many times. Google’s surge was due to then-search giants looking for a way to generate enough revenue to pay for the cost of indexing content. Now there are some crazy ideas floating around “real” consultant cubicles that search is cheap and easy.
Next Generation Portals: Gate to Revenue Hell?
Fear not, lads and lasses.
Search is brutally expensive and—guess what?—the costs keep on rising. The principal reasons are that systems need constant mothering and money. No, let’s not forget the costs which MBAs often trivialize. These include people who can make the system work, run faster, remain online, and keep pace with technology. Telecommunications, power, hardware, and a number of other “trivial” items gobble money like a 12 year old soccer player after practice chowing down on junk food.
Next Generation Portals: Gate to Revenue Heaven?
Portals promised to be “sticky”, work like magnets and pull more users, and provide a platform for advertising. Portals were supposed to make money when search generated modest amounts of money. First Overture, then Yahoo, and finally Google realized that the pursuit of objectivity was detrimental to selling traffic. Thus, online pay-to-play programs took off. The portals with a lead like Yahoo fumbled the ball. The more clever Googlers grabbed the melon and kept going back to the farmer’s garden for more. Google had, it appeared, figured out how to remain a search system and make lots of money.
No more.
Do we now witness the portalization of Google? Is the new twist is that the Google portal will require the user to have multiple devices? Will each device connects to Google to show more portal advertising goodness?
There is a popular impression among some MBAs on Wall Street and “real” consultants that Google is riding the same old money rocket in did in 2004 to 2006. My view is different.
Amazon and Pinterest: Information or Disinformation
August 30, 2012
I found “Uh Oh! Amazon Researchers Say Pinterest Doesn’t Generate A Lot Of Sales” interesting. Why? When a major Web outfit does research which suggests another Web outfit is either good or bad, my radar pings. Is the article praising Pinterest or is the article criticizing Pinterest. I find survey data remarkably malleable. Information can be shaped in different ways.
Consider this statement:
Will Young, director of Zappos Labs, told Bloomberg that Pinterest users are far more likely to share a purchase than Twitter or Facebook users—but that shared items generate far less revenue than Twitter or Facebook. This is a big problem for Pinterest, because the whole idea of the site is that it’s supposed to be better at monetizing social activity than Twitter or Facebook.
We have a positive statement about the sharing and a negative statement about generating sales.
You decide. My questions are:
- Is Amazon planning on competing with or buying Pinterest? A touch of doubt is useful when negotiating.
- Does it matter to Pinterest users if they don’t buy at a rate another retail site thinks is “low”? If Pinterest users are happy, does survey data matter?
- Were the survey data assembled doing the “right stuff.” When creating questions and analyzing survey data, tiny choices can have major implications for the outputs. I am confident that Amazon Zappos worked hard and did the maths correctly. I am just asking, “Was the finding one that required some tiny decisions about which I know nothing?”
Pinterest, in my opinion, works for a specific demographic segment. “Works” does not mean buy Mr Clean bleach pens on Amazon or snap up discounted evening slippers on Zappos.
Search is subject to similar tiny decisions. What happens when I try just published books on Amazon? I get pages of results listing books that will be published in the future. Tiny decision? Carelessness? An attempt to make a sale before having the product?
Stephen E Arnold, August 30, 2012
Sponsored by Augmentext
Semantics and Facets in Mindbreeze InSite
August 30, 2012
Fabasoft Mindbreeze will not only meet your internal enterprise search needs, but can also meet external needs in terms of your public facing Web site. Fabasoft Mindbreeze InSite can provide your users with intuitive search on your Web presence. The Fabsoft Mindbreeze blog entry, “Semantics and Facets: Mindbreeze InSite Makes Your News Cosmos ‘Sexy,’” pays attention to the topic.
The author begins:
The key term is semantic search, which means ‘understanding content.’ And I’m proud to be able to report that Mindbreeze InSite can do this perfectly. But how does it work? It’s not magic code, just statistical mathematics and sophisticated content analysis. Both aspects molded together in one program give Mindbreeze its power to ‘think.’ Through content analysis and statistics, Mindbreeze InSite really can be tailored to meet the individual needs of each website operator. Essentially the result is always the same: The perfect search results you need quickly. You can also search, for example, according to author, topic, tags and terms.
The semantic and faceted search can pay real dividends for your business. Users appreciate a hassle-free and intuitive experience when navigating a Web site. This can be especially useful in retail contexts. In addition, Fabasoft Mindbreeze backs all their products with outstanding customer support and frequent updates.
Emily Rae Aldridge, August 30, 2012
Sponsored by ArnoldIT.com, developer of Augmentext.
WAND Inc Makes Integration Plans with Nintex Workflow Official
August 30, 2012
Last week WAND, Inc announced its plans to integrate with Nintex Workflow. Wand, Inc is attributed to making search work better by the use of its various taxonomies and Nintex is the world’s leader in SharePoint Workflow. Separately they are powerhouses in their specific field and combined they seem to have a lot to offer each other. The article, WAND adds automatic tagging to Nintex Workflow sheds some more light on what this match up could mean:
“The DataFacet Automatic Annotation custom action for Nintex Workflow allows a user to automatically tag documents with taxonomy metadata as part of a workflow process. Users who have DataFacet and Nintex Workflow …will be able take advantage of this custom action to control when a document is automatically tagged and base conditional actions on those tags. Workflows can be configured to tag documents with any terminology that is stored in the SharePoint Term Store, including terms from any of WAND’s Foundation Taxonomies.”
If everything runs smoothly for these two, this seems like a decent match up that has some great potential. I think it is safe to say that we can be expecting some great things from these guys in the coming years.
Edie Marie, August 30, 2012
Sponsored by ArnoldIT.com, developer of Augmentext
-
- Subscribe to Beyond Search
Feature archive
News archive
-
