Are Auto Suggestions Inherently Problematic?
June 3, 2018
Politics is a dangerous subject to bring up in any social situation. My advice is to keep quiet and nod, then you can avoid loudmouths trying to press their agendas down your throat. Despite attempts to remain polite, the Internet always brings out the worst in people and The Sun shares how with a simple search engine function, “‘Trump Should Be Shot’ Google And Bing Searches For ‘Trump’ And ‘Conservatives’ Offer Disgusting Auto-Suggestions.”
Auto-complete is notorious for making hilarious mistakes and the same is with auto-suggest on search engines, but these end up to be more gruesome than a misspelling.
If you want to see some interesting suggestions, type “Trump should be…” into a blank search bar and the results are endless, including: shot, arrested, killed, in jail, arrested banned from Twitter (okay, the last one might be a little funny).
Typing in “conservatives need…” results in less derogatory terms, but the auto-suggestions include: to die, to go, a new party, and not apply.
Hmmm.
What creates these auto-suggestions?
“These are based on a number of factors including real-time searches, trending results, your location, and previous activity.The intuitive predictions change in “response to new characters being entered into the search box” explains Google. And the company also has its own set of “autocomplete policies” in case something untoward should pop up.Along with prohibiting predictions that contain sexually explicit, violent, and harmful terms, Google says it also removes hateful suggestions against groups and individuals. ‘We remove predictions that include graphic descriptions of violence or advocate violence generally,’ states the firm.”
Google and Bing deserve some credit for removing the slander from auto-complete, but sometimes they only do it when they are pushed. Trolls and bigots create these terms and it would be nice to see them scrubbed from auto-suggest, but it is near impossible. Hey, Bing and Google try scrubbing 4chan!
Whitney Grace, June 3, 2018
Google: Excellence Evolves to Good Enough
May 25, 2018
I read “YouTube’s Infamous Algorithm Is Now Breaking the Subscription Feed.” I assume the write up is accurate. I believe everything I read on the Internet.
The main point of the write up seems to me to be that good enough is the high water mark.
I noted this passage, allegedly output by a real, thinking Googler:
Just to clarify. We are currently experimenting with how to show content in the subs feed. We find that some viewers are able to more easily find the videos they want to watch when we order the subs feed in a personalized order vs always showing most recent video first.
I also found this statement interesting:
With chronological view thrown out, it’s going to become even more difficult to find new videos you haven’t seen — especially if you follow someone who uploads at a regular time each day.
I would like to mention that Google, along wit In-Q-Tel, invested in Recorded Future. That company has some pretty solid date and time stamping capabilities. Furthermore, my hunch is that the founders of the company know the importance of time metadata to some of the Recorded Future customers.
What would happen if Google integrated some of Recorded Future’s time capabilities into YouTube and into good old Google search results.
From my point of view, good enough means “sells ads.” But I am usually incorrect, and I expect to learn just how off base I am when I explain how one eCommerce giant is about to modify the landscape for industrial strength content analysis. Oh, that company’s technology does the date and time metadata pretty well.
More on this mythical “revolution” on June 5th and June 6th. In the meantime, try and find live feeds of the Hawaii volcano event using YouTube search. Helpful, no?
Stephen E Arnold, May 25, 2018
LightTag Helps AI Developers Label Training Data
May 16, 2018
The creators of LightTag are betting on the AI boom, we learn from TechCrunch’s post, “LightTag Is a Text Annotation Platform for Data Scientists Creating AI Training Data.” Built by a former Natural Language researcher for Citigroup, the shiny new startup hopes to assist AI developers with one of their most labor-intensive and error-prone tasks—labeling the data used to train AI systems. Since it is a job carried out by teams of imperfect humans, errors often abound. LightTag’s team-based workflow, user interface, and quality controls are designed to mitigate these imperfections. Writer Steve O’Hear cites founder Tal Perry as he reports:
“Perry says LightTag’s annotation interface is designed to keep labelers ‘effective and engaged’. It also employs its own ‘AI’ to learn from previous labeling and make annotation suggestions. The platform also automates the work of managing a project, in terms of assigning tasks to labelers and making sure there is enough overlap and duplication to keep accuracy and consistency high. ‘We’ve made it dead-simple to annotate with a team (sounds obvious, but nothing else makes it easy),’ he says. ‘To make sure the data is good, LightTag automatically assigns work to team members so that there is overlap between them. This allows project managers to measure agreement and recognize problems in their project early on. For example, if a specific annotator is performing worse than others’.”
For the organizations in certain industries like healthcare, law, and banking that simply cannot risk outsourcing the task, LightTag offers an on-premise version. The write-up includes a couple GIFs of the software at work, so check it out if curious. Though it only recently launched publicly, the beta software has been tried out by select clients, including these noteworthy uses: An energy company is using it to predict drilling issues at certain depths with data from oil-rig logs, and a medical imaging company has used it to label MRI-scan reports. We are curious to see whether the young startup will be able to capitalize on the current AI boom, as Perry predicts.
Cynthia Murrell, May 16, 2018
Free Keyword Research Tools
May 15, 2018
Short honk: Search Engine Watch published a write up intended for SEO experts. The article contained some useful links to free keyword search tools. Even if you are not buying online ads or fiddling with your indexing, the services are interesting to know about. Here they are:
- Bulk Keyword at http://tools.seochat.com/tools/suggest-tool/#sthash.Jbkkmbfm.dpbs
- Google Correlate at https://www.google.com/trends/correlate
- Keyword.Guru at http://keyword.guru/
- Serpstat at https://serpstat.com/
- Soovle at https://soovle.com/
- Ubersuggest at https://neilpatel.com/ubersuggest/
- Wordstream at https://www.wordstream.com/keywords/
Stephen E Arnold, May 15, 2018
Google Search and Hot News: Sensitivity and Relevance
November 10, 2017
I read “Google Is Surfacing Texas Shooter Misinformation in Search Results — Thanks Also to Twitter.” What struck me about the article was the headline; specifically, the implication for me was that Google was not responding to user queries. Google is actively “surfacing” or fetching and displaying information about the event. Twitter is also involved. I don’t think of Twitter as much more than a party line. One can look up keywords or see a stream of content containing a keyword or a, to use Twitter speak, “hash tags.”
The write up explains:
Users of Google’s search engine who conduct internet searches for queries such as “who is Devin Patrick Kelley?” — or just do a simple search for his name — can be exposed to tweets claiming the shooter was a Muslim convert; or a member of Antifa; or a Democrat supporter…
I think I understand. A user inputs a term and Google’s system matches the user’s query to the content in the Google index. Google maintains many indexes, despite its assertion that it is a “universal search engine.” One has to search across different Google services and their indexes to build up a mosaic of what Google has indexed about a topic; for example, blogs, news, the general index, maps, finance, etc.
Developing a composite view of what Google has indexed takes time and patience. The results may vary depending on whether the user is logged in, searching from a particular geographic location, or has enabled or disabled certain behind the scenes functions for the Google system.
The write up contains this statement:
Safe to say, the algorithmic architecture that underpins so much of the content internet users are exposed to via tech giants’ mega platforms continues to enable lies to run far faster than truth online by favoring flaming nonsense (and/or flagrant calumny) over more robustly sourced information.
From my point of view, the ability to figure out what influences Google’s search results requires significant effort, numerous test queries, and recognition that Google search now balances on two pogo sticks. Once “pogo stick” is blunt force keyword search. When content is indexed, terms are plucked from source documents. The system may or may not assign additional index terms to the document; for example, geographic or time stamps.
The other “pogo stick” is discovery and assignment of metadata. I have explained some of the optional tags which Google may or may not include when processing a content object; for example, see the work of Dr. Alon Halevy and Dr. Ramanathan Guha.
But Google, like other smart content processing today, has a certain sensitivity. This means that streams of content processed may contain certain keywords.
When “news” takes place, the flood of content allows smart indexing systems to identify a “hot topic.” The test queries we ran for my monographs “The Google Legacy” and “Google Version 2.0” suggest that Google is sensitive to certain “triggers” in content. Feedback can be useful; it can also cause smart software to wobble a bit.
T shirts are easy; search is hard.
I believe that the challenge Google faces is similar to the problem Bing and Yandex are exploring as well; that is, certain numerical recipes can over react to certain inputs. These over reactions may increase the difficulty of determining what content object is “correct,” “factual,” or “verifiable.”
Expecting a free search system, regardless of its owner, to know what’s true and what’s false is understandable. In my opinion, making this type of determination with today’s technology, system limitations, and content analysis methods is impossible.
In short, the burden of figuring out what’s right and what’s not correct falls on the user, not exclusively on the search engine. Users, on the other hand, may not want the “objective” reality. Search vendors want traffic and want to generate revenue. Algorithms want nothing.
Mix these three elements and one takes a step closer to understanding that search and retrieval is not the slam dunk some folks would have me believe. In fact, the sensitivity of content processing systems to comparatively small inputs requires more discussion. Perhaps that type of information will come out of discussions about how best to deal with fake news and related topics in the context of today’s information retrieval environment.
Free search? Think about that too.
Stephen E Arnold, November 10, 2017
Smartlogic: A Buzzword Blizzard
August 2, 2017
I read “Semantic Enhancement Server.” Interesting stuff. The technology struck me as a cross between indexing, good old enterprise search, and assorted technologies. Individuals who are shopping for an automatic indexing systems (either with expensive, time consuming hand coded rules or a more Autonomy-like automatic approach) will want to kick the tires of the Smartlogic system. In addition to the echoes of the SchemaLogic approach, I noted a Thomson submachine gun firing buzzwords; for example:
best bets (I’m feeling lucky?)
dynamic summaries (like Island Software’s approach in the 1990s)
faceted search (hello, Endeca?)
model
navigator (like the Siderean “navigator”?)
real time
related topics (clustering like Vivisimo’s)
semantic (of course)
taxonomy
topic maps
topic pages (a Google report as described in US29970198481)
topic path browser (aka breadcrumbs?)
visualization
What struck me after I compiled this list about a system that “drives exceptional user search experiences” was that Smartlogic is repeating the marketing approach of traditional vendors of enterprise search. The marketing lingo and “one size fits all” triggered thoughts of Convera, Delphes, Entopia, Fast Search & Transfer, and Siderean Software, among others.
I asked myself:
Is it possible for one company’s software to perform such a remarkable array of functions in a way that is easy to implement, affordable, and scalable? There are industrial strength systems which perform many of these functions. Examples range from BAE’s intelligence system to the Palantir Gotham platform.
My hypothesis is that Smartlogic might struggle to process a real time flow of WhatsApp messages, YouTube content, and mobile phone intercept voice calls. Toss in the multi language content which is becoming increasingly important to enterprises, and the notional balloon I am floating says, “Generating buzzwords and associated over inflated expectations is really easy. Delivering high accuracy, affordable, and scalable content processing is a bit more difficult.”
Perhaps Smartlogic has cracked the content processing equivalent of the Voynich manuscript.
Will buzzwords crack the Voynich manuscript’s inscrutable text? What if Voynich is a fake? How will modern content processing systems deal with this type of content? Running some content processing tests might provide some insight into systems which possess Watson-esque capabilities.
What happened to those vendors like Convera, Delphes, Entopia, Fast Search & Transfer, and Siderean Software, among others? (Free profiles of these companies are available at www.xenky.com/vendor-profiles.) Oh, that’s right. The reality of the marketplace did not match the companies’ assertions about technology. Investors and licensees of some of these systems were able to survive the buzzword blizzard. Some became the digital equivalent of Ötzi, 5,300 year old iceman.
Stephen E Arnold, August 2, 2017
AI Not to Replace Lawyers, Not Yet
May 9, 2017
Robot or AI lawyers may be effective in locating relevant cases for references, but they are far away from replacing lawyers, who still need to go to the court and represent a client.
ReadWrite in a recently published analytical article titled Look at All the Amazing Things AI Can (and Can’t yet) Do for Lawyers says:
Even if AI can scan documents and predict which ones will be relevant to a legal case, other tasks such as actually advising a client or appearing in court cannot currently be performed by computers.
The author further explains that what the present generation of AI tools or robots does. They merely find relevant cases based on indexing and keywords, which was a time-consuming and cumbersome process. Thus, what robots do is eliminate the tedious work that was performed by interns or lower level employees. Lawyers still need to collect evidence, prepare the case and argue in the court to win a case. The robots are coming, but only for doing lower level jobs and not to snatch them.
Vishol Ingole, May 9, 2017
Palantir Technologies: A Beatdown Buzz Ringing in My Ears
April 27, 2017
I have zero contacts at Palantir Technologies. The one time I valiantly contacted the company about a speaking opportunity at one of my wonky DC invitation-only conferences, a lawyer from Palantir referred my inquiry to a millennial who had a one word vocabulary, “No.”
There you go.
I have written about Palantir Technologies because I used to be an adviser to the pre-IBM incarnation of i2 and its widely used investigation tool, Analyst’s Notebook. I did write about a misadventure between i2 Group and Palantir Technologies, but no one paid much attention to my commentary.
An outfit called Buzzfeed, however, does pay attention to Palantir Technologies. My hunch is that the online real news outfit believes there is a story in the low profile, Peter Thiel-supported company. The technology Palantir has crafted is not that different from the Analyst’s Notebook, Centrifuge Systems’ solution, and quite a few other companies which provide industrial-strength software and systems to law enforcement, security firms, and the intelligence community. (I list about 15 of these companies in my forthcoming “Dark Web Notebook.” No, I won’t provide that list in this free blog. I may be retired, but I am not giving away high value information.)
So what’s caught my attention. I read the article “Palantir’s Relationship with the Intelligence Community Has Been Worse Than You Think.” The main idea is that the procurement of Palantir’s Gotham and supporting services provided by outfits specializing in Palantir systems has not been sliding on President Reagan’s type of Teflon. The story has been picked up and recycled by several “real” news outfits; for example, Brainsock. The story meshes like matryoshkas with other write ups; for example, “Inside Palantir, Silicon Valley’s Most Secretive Company” and “Palantir Struggles to Retain Clients and Staff, BuzzFeed Reports.” Palantir, it seems to me in Harrod’s Creek, is a newsy magnet.
The write up about Palantir’s lousy relationship with the intelligence community pivots on a two year old video. I learned that the Big Dog at Palantir, Alex Karp, said in a non public meeting which some clever Hobbit type videoed on a smartphone words presented this way by the real news outfit:
The private remarks, made during a staff meeting, are at odds with a carefully crafted public image that has helped Palantir secure a $20 billion valuation and win business from a long list of corporations, nonprofits, and governments around the world. “As many of you know, the SSDA’s recalcitrant,” Karp, using a Palantir codename for the CIA, said in the August 2015 meeting. “And we’ve walked away, or they walked away from us, at the NSA. Either way, I’m happy about that.” The CIA, he said, “may not like us. Well, when the whole world is using Palantir they can still not like us. They’ll have no choice.” Suggesting that the Federal Bureau of Investigation had also had friction with Palantir, he continued, “That’s de facto how we got the FBI, and every other recalcitrant place.”
Okay, I don’t know the context of the remarks. It does strike me that 2015 was more than a year ago. In the zippy doo world of Sillycon Valley, quite a bit can change in one year.
I don’t know if you recall Paul Doscher who was the CEO of Exalead USA and Lucid Imagination (before the company asserted that its technology actually “works). Mr. Doscher is a good speaker, but he delivered a talk in 2009, captured on video, during which he was interviewed by a fellow in a blue sport coat and shirt. Mr. Doscher wore a baseball cap in gangsta style, a crinkled unbuttoned shirt, and evidenced a hipster approach to discussing travel. Now if you know Mr. Doscher, he is not a manager influenced by gangsta style. My hunch is that he responded to an occasion, and he elected to approach travel with a bit of insouciance.
Could Mr. Karp, the focal point of the lousy relationship article, have been responding to an occasion? Could Mr. Karp have adopted a particular tone and style to express frustration with US government procurement? Keep in mind that a year later, Palantir sued the US Army. My hunch is that views expressed in front of a group of employees may not be news of the moment. Interesting? Sure.
What I find interesting is that the coverage of Palantir Technologies does not dig into the parts of the company which I find most significant. To illustrate: Palantir has a system and method for an authorized user to add new content to the Gotham system. The approach makes it possible to generate an audit trail to make it easy (maybe trivial) to answer these questions:
- What data were added?
- When were the data added?
- What person added the data?
- What index terms were added to the data?
- What entities were added to the metadata?
- What special terms or geographic locations were added to the data?
You get the idea. Palantir’s Gotham brings to intelligence analysis the type of audit trail I found some compelling in the Clearwell system and other legal oriented systems. Instead of a person in information technology saying in response to a question like “Where did this information come from?”, “Duh. I don’t know.”
Gotham gets me an answer.
For me, explaining the reasoning behind Palantir’s approach warrants a write up. I think quite a few people struggling with problems of data quality and what is called by the horrid term “governance” would find Palantir’s approach of some interest.
Now do I care about Palantir? Nah.
Do I care about bashing Palantir? Nah.
What I do care about is tabloidism taking precedence over substantive technical approaches. From my hollow in rural Kentucky, I see folks looking for “sort of” information.
How about more substantive information? I am fed up with podcasts which recycle old information with fake good cheer. I am weary of leaks. I want to know about Palantir’s approach to search and content processing and have its systems and methods compared to what its direct competitors purport to do.
Yeah, I know this is difficult to do. But nothing worthwhile comes easy, right?
I can hear the millennials shouting, “Wrong, you dinosaur.” Hey, no problem. I own a house. I don’t need tabloidism. I have picked out a rest home, and I own 60 cemetery plots.
Do your thing, dudes and dudettes of “real” journalism.
Stephen E Arnold, April 27, 2017
Palantir Technologies: 9000 Words about a Secretive Company
April 3, 2017
Palantir Technologies is a search and content processing company. The technology is pretty good. The company’s marketing pretty good. Its public profile is now darned good. I don’t have much to say about Palantir’s wheel interface, its patents, or its usefulness to “operators.” If you are not familiar with the company, you may want to read or at least skim the weirdo Fortune Magazine Web article “Donald Trump, Palantir, and the Crazy Battle to Clean Up a Multibillion Dollar Military Procurement Swamp.” The subtitle is a helpful statement:
Peter Thiel’s software company says it has a product that will save soldiers’ lives—and hundreds of millions in taxpayer funds. The Army, which has spent billions on a failed alternative, isn’t interested. Weill the president and his generals ride to the rescue?”
The article, minus the pull quotes, is more than 9000 words long. The net net of the write up is that changing the US government’s method of purchasing goods and services may be tough to modify. I used to work at a Beltway Bandit outfit. Legend has it that my employer helped set up the US Department of the Navy and many of the business processes so many contractors know and love.
One has to change elected officials, government professionals who operate procurement processes, outfits like Beltway Bandits, and assorted legal eagles.
Why take 9000 words to reach this conclusion. My hunch is that the journey was fun: Fun for the Fortune Magazine staff, fun for the author, and fun for the ad sales person who peppered the infinite page with ads.
Will Palantir Technologies enjoy the write up? I suppose it depends on whom one asks. Perhaps a reader connected to IBM could ask Watson about the Analyst’s Notebook team. What are their views of Palantir? For most folks, my thought is that the Palantir connection to President Trump may provide a viewshed from which to assess the impact of this real journalism essay thing.
Stephen E Arnold, April 3, 2017
Is Google Plucking a Chicken Joint?
March 14, 2017
Real chicken or fake news? You decide. I read “Google, What the H&%)? Search Giant Wrongly Said Shop Closed Down, Refused to List the Truth.” The write up reports that a chicken restaurant is clucking mad about how Google references the eatery. The Google, according to the article, thinks the fowl peddler is out of business. The purveyor of poultry disagrees.
The write up reports:
Kaie Wellman says that her rotisserie chicken outlet Arrosto, in Portland, Oregon, US, was showing up as “permanently closed” on Google’s mobile search results.
Ms Wellman contacted the Google and allegedly learned that Google would not change the listing. The fix seems to be that the bird roaster has to get humans to input data via Google Maps. The smart Google system will recognize the inputs and make the fix.
The write up reports that the Google listing is now correct. The fowl mix up is now resolved.
Yes, the Google. Relevance, precision, recall, and accuracy. Well, maybe not so much of these ingredients when one is making fried mobile outputs.
Stephen E Arnold, March 14, 2017
-
- Subscribe to Beyond Search
Feature archive
News archive
-
