A Googler on Prediction Markets
June 20, 2010
One of Endeca’s wizards jumped to Google. I have not paid much attention to what former Endeca Googlers do. The machinations of the Google are too far removed from the goose pond here in Harrod’s Creek. However, a reader sent me a link to the story “Why Can’t We Just Use Prediction Markets?” I read the article and did not find much with which to disagree. However, I asked myself, “Is this write up about prediction markets a personal opinion or one of those moments when a Googler sends a signal (intentionally or unintentionally) about notions that fascinate the users of MOMA.
For the fun of it, let’s assume that the article does reveal a fragment of Google’s thinking about the importance and utility of prediction markets. As you may know, a prediction market takes guesses and tries to figure out the future. There are many fancy ways to explain the method, but the idea is that some unidentified, latent magic exists when people offer opinions or bet on something. The data provide a glimpse of the future. I wish to point out that the US government has been known to think about prediction markets, and we know how well most US government decisions work out. Here in Harrod’s Creek we don’t need a predication market to figure out that there will be some unhappy people where the BP “spill” adds to the salty sea air zest.
Three items warrant comment for this alternative view of the write up.
First, the article suggests reviews (Amazon, are your with me here) can be manipulated by humans. The notion of a prediction method might generate a more useful indicator about a product. I think numerical recipes processing Big Data can yield some useful outputs. Google could make a run at Amazon and other review sites, right?
Second, the notion of using data to bet that I would like something also makes sense. I can see the Google kicking the notion of reviews into recommendation land. Again, I am thinking of Amazon and its recommendations. Maybe Google will be doing recommendations in its forthcoming Google TV service. What do you think about the Google predicting what rich media will make you a happy buyer or camper?
Finally, the concept of accuracy is delightful. Now accuracy in the Math Club works out to the old six nines confidence level. But accuracy also means a confidence score better than the lousy confidence score generated by other systems. If Google can generate a higher confidence score, that will give the company an edge. We are not talking about perfection. We are talking about “better than”, right?
My take.
Google will be rolling out more recommendations and predictive services. You may not use them directly, but these methods will allow the company to suggest that if you liked the NBC TV show about Google, you will definitely love these other TV shows. No humans needed to publish this type of indirect Consumers Report information. Many other uses as well I assert.
Now that’s my take away from the article.
Stephen E Arnold, June 20, 2010
Freebie
Quote to Note: Fix the Audit
June 20, 2010
This quote is not about search or content processing. In my opinion it is one of the most remarkable statements I have read in a quite a while. Navigate to “KPMG Chief Calls for Audit Reform” on the Accountancy Age Web site. Yep, accountants: The same fine certified group of professionals who have done so much for shareholders and employees of Tyco, Enron, et al. We may as well toss in the Securities & Exchange Commission, another outfit with legions of these professionals. Now here is the quote from KPMG’s top bean counter, John Griffith hyphen Jones. The hyphen adds class to a profession already replete with upscale virtues:
“What is the point, they and others ask, of doing extensive and increasingly elaborate audits of the financial accounts of our banks, when audits failed to identify the huge and systemic risks which led to the near collapse of the Global banking system in the Autumn of 2008?” he said. “It is a straightforward question; It deserves a straightforward answer.”
Yep, straightforward. For some local color, read “Bailed-Out Banker Praises Capitalism, Attacks Parasites.” My question: Are accountants parasites?
Stephen E Arnold, June 20, 2010
Freebie
Semantic Search Explained
June 19, 2010
I get asked about semantic search one a day, often more frequently. I usually say, “Semantic search means software can figure out what something is about.” If that does not do the trick, I trot out the more detailed explanation Martin White and I put in our 2009 study “Successful Enterprise Search Management.”
I neglected to write about “10 Things that Make Search a Semantic Search.” The informton in that write up by the founder of Hakia, Dr. Riza C. Berkan is useful. If you have not reviewed the write up, you will want to put this reading on your To Do list.
I don’t want to reproduce the full list. Navigate to the original article and work through. I do want to highlight three points with which I agree.
First, a semantic search can handle synonyms. Languages are like roads in Kentucky, full of potholes. Disambiguation and figuring out synonyms are two important tasks. Their presence signals a semantic component in the content processing system.
Second, a search systm that can present a snippet or a highlight of the key sentence of paragraph is quite useful. I find that some snippeting technology is designed to meet the needs of folks selling ads. The snippeting function I want works with the honesty and zeal of a prisoner who is due to be released from prison in two days.
Finally, a user can enter a query without having to formulate a query with Boolean operators or special instructions such as CC=. Systems have to be smart but not biased or tilted for the benefit of advertisers. Objectivity is important in delivering this type of query support. Alas, I think this is a difficult goal to achieve. Humans are humans and often prefer to click the ad for a vacation rental than running a query and perusing results, then making an informed decision.
A happy quack to Hakia for the post.
Stephen E Arnold, June 19, 2010
Freebie
Oracle and SAP Chase Big Data Rainbow
June 19, 2010
“Oracle, SAP Working on Exadata Support” struck me as interesting for three reasons. First, if you are managing one of these super scale storage gizmos, you have some challenges on your hands. Second, Oracle and SAP know that leasers’ of these storage devices have those problems and aim to cash in on the situation. And, third, the outfit that figures out how to make these gizmos work will have bragging rights in the hyper-expensive, enterprise storage market.
High stakes indeed.
Why are big data a problem? Many reasons. The obvious one is that big data take “time” to transform, manipulate, and crunch. The good news is that the problem can be solved by buying more Exadata database machines. Better yet, SAP wants you to buy a Sybase gizmo. The bad news is that adding machines creates more management hassles for the engineers. The less obvious one is that an Exadata gizmo is not one of the slick No SQL solutions that rely on lower cost methods. The good news is that a Fortune 100 company may not trust No SQL or not know much about a No SQL solution. The bad news is that today’s Fortune 100 company could become tomorrow’s employment grave yard. Quiet places, graveyards. The spat between the two companies is not interesting to me. Squabbles that most people don’t understand are good for the azure chip crowd and bloggers. Regular folks, not so much.
The most interesting comment in the write up was:
SAP is now supporting Oracle Database 11g R2, for applications that use SAP kernel 6.40, 7.x and beyond. The companies’ practice has been to delay certifying Oracle’s database releases until the second iteration, a process that minimizes upgrade chores for customers.
This is a service-for-a-fee game. I am not sure the needs of the customer are front and center. The gizmos, the complexity, and the support are the main event.
Stephen E Arnold, June 19, 2010
Freebie
Mashable Seems to Highlight Weaknesses in Google Search
June 19, 2010
Short honk: I have a twitch in my goose brain that says, Mashable wanted to help Android phone users find applications. The write up spoke to me a different way. I read “7 Ways to Find Amazing New Android Apps” and realized that Google’s own search system is not delivering the goods. Maybe Google is distracted? Maybe Google’s Math Club perceives native Android app search as the omega. Read the Mashable story and decide for yourself. Excellent search, it seems, requires one to use multiple services. How about a metasearch system?
Stephen E Arnold, June 8, 2010
Freebie
Another Googler Feeds the PR Blast Furnace
June 18, 2010
Navigate to “Working at Google – The First 6 Months.” Sound familiar? Consider these words and phrases:
- Start up
- Engineering rules
- Change quickly
- Work hard, have fun
- Hiring now.
I have some questions for you, gentle reader:
- Is “start up” and any connotations are universal “good thing”? We have some products that did not just disappoint, the products revealed the nature of Google’s ability to hook action and engineering to customer needs. Examples include Google Buzz, Google Wave, and assorted YAGGs (yet another Google glitch).
- Engineering can be a good thing, but as the Nexus One “customer support” method demonstrated, humans want more customer support than a chipper email and a link to Google’s next to impossible thicket of Web pages, documents, and code snippets.
- Change quickly can be a virtue; however, the notion of pinning the Wi Fi data collection on a “rogue engineer” and triggering some legal excitement in the US and elsewhere suggests that more may be required
- Work hard, have fun does have a nice ring to it, but there are lots of Xooglers running around now, and I am not sure some of these folks have the appetite for the Kool-Aid despite their compensation, free lunches, and serenades by Tony Benett. (Oh, Mr. Bennett lose that autoplay music, please.)
Each high profile hire at Google rolls out a PR encomium: Bray, Tunkenlang, et al. The addled goose expects more from a 10 year old. Come to think of it so do legal authorities in Germany, bureaucrats in China and Vietnam, and partners like HTC who must cope with new guidelines for the Android interface. Just my opinion. Honk.
Stephen E Arnold, June 18, 2010
EasyAsk eCommerce Mobile
June 18, 2010
“EasyAsk Introduces Easy Ask eCommerce Mobile Edition” revealed that the search vendor has gone mobile. EasyAsk describes itself as an industry leading provider of natural language search and analysis software for ecommerce Web sites. The Mobile Edition delivers such key features as:
- Customized mobile web pages, integrating ecommerce search, navigation and merchandising directly into these pages. Mobile visitors are automatically detected and routed to the mobile version of the site, delivering a unique experience.
- An optimized experience to the specific handset by detecting the model of the phone, matching it with a database of thousands of handsets, and uniquely optimizing the user experience to the capability of the consumer handset.
- Integration with the services from the main website for cross promotions, such as finding specific items in the brick-and-mortar store where customers can touch and feel products. It also enables delivery of discount or cross-sell offers to the phone, enticing immediate, and even larger, in-store purchases.
The Mobile Edition taps the EasyAsk Commerce Studio so licensees can tailor product presentation, offers and other merchandising attributes. The Mobile Edition provides analytics to provide insight into customer behavior.
You can get more information about EasyAsk at www.easyask.com. A quick search of the Progress Software Web site did not return information about EasyAsk eCommerce Mobile Edition. Interesting.
Stephen E Arnold, June 18, 2010
Freebie
Social Networks, Bigger than Search?
June 18, 2010
Short honk: Surf to “Social Networks More popular than Search Engines.” Note the tiny little red line that crosses the tiny little blue line. No real data to back up the itty bitty intersection with the tiny red poke upwards. But, if the graph is accurate, maybe search has been eclipsed by social networks. Here’s the passage I found interesting:
More people visited social-networking sites than used search engines in May, according to web-analysis firm Hitwise. It is the first time that social networking has been more popular than web search, according to the company.
Probably a fluke. Oh, Bank of America downgraded the Google on June 8, 2010. Probably one of those coincidences.
Stephen E Arnold, June 18, 2010
Freebie
Some Tips for Oracle Text Users
June 18, 2010
Lucky you! Your task is to use Oracle Text to search content in your Oracle Database. Oracle provides some documentation, and you can spend many pleasant hours chasing down the documentation and specific information you need to make Oracle Text do your bidding. If you are short on time, point your browser at “Setting Up Oracle Database’s Oracle Text” by Steven Callan. Mr. Callan does a good job of explaining how to get from A to B. The most significant parts of the write up in my opinion are the examples. You can learn where Oracle tucks the search system when you install the Oracle database. He provides specific information for figuring the schema of the table. Of particular value is the code example for using different index types. Recommended. If you want to crack the performance and index update problems, you may have to do some digging on your own.
Stephen E Arnold, June 18, 2010
Freebie
A Smidgen of Dataspace Seeps from the Google
June 18, 2010
My favorite dataspace guru and his colleagues have nailed a US patent. I am speaking of Dr. Alon Halevy and his clutch of merry Googlers, Jayant Madhavan and David Ko. I know the azure chip set is excited about Google’s ability to pinpoint gaps in media coverage and methods for sucking the filling out of Wi Fi content. I am not. Nope, I keep my eye on the ball. In this case the system and method disclosed in US7,739,258, “Facilitating Searches through Content Which Is Accessible through Web-Based Forms.” These are not your mother’s social security forms. With some prior patents in the possession of sporty outfits like AT&T and Lucent, the Googlers have to tiptoe through the dataspace. I was going to write a lyric for Tiny Tim to sing: something along the line of “Tiptoe through the manifold with me.” But I won’t.
Here’s the crystal clear prose from the Google wizard and legal eagle team:
One embodiment of the present invention provides a system that facilitates crawling through web-based forms to gather information to facilitate subsequent searches through content which is accessible though the web-based forms. During operation, the system first obtains web-based forms to be searched. Note that the system can obtain these web-based forms from a number of sources. For example, the system can crawl through web sites to identify web-based forms, the system can receive manually provided web-based forms, or the system can find web-based forms through methods other than crawling. Next, the system creates database entries for the identified forms. This involves obtaining and storing metadata describing the identified forms into database entries and then storing these database entries in a form database to facilitate searches through content which is accessible through the identified forms. Note that this form database can include a web index and associated documents, which can be used to facilitate web search queries that return both ordinary documents and documents that result from form queries.
My view? Important.
Stephen E Arnold, June 18, 2010
Freebie
-
- Subscribe to Beyond Search
Feature archive
News archive
-
