Reengineering Bias: What an Interesting Idea

April 5, 2021

If this is true, AI may be in trouble now. VentureBeat reports, “Researchers Find that Debiasing Doesn’t Eliminate Racism from Hate Speech Detection Models.” It is known that AI systems meant to detect toxic language themselves have a problem with bias. Specifically, they tend to flag text by Black users more often than text by white users. Oh, the irony. The AI gets hung up on language markers often found in vernaculars like African-American English (AAE). See the article for a few examples. Researchers at the Allen Institute tried several techniques to reteach existing systems to be more even handed. Reporter Kyle Wiggers writes:

“In the course of their work, the researchers looked at one debiasing method designed to tackle ‘predefined biases’ (e.g., lexical and dialectal). They also explored a process that filters ‘easy’ training examples with correlations that might mislead a hate speech detection model. According to the researchers, both approaches face challenges in mitigating biases from a model trained on a biased dataset for toxic language detection. In their experiments, while filtering reduced bias in the data, models trained on filtered datasets still picked up lexical and dialectal biases. Even ‘debiased’ models disproportionately flagged text in certain snippets as toxic. Perhaps more discouragingly, mitigating dialectal bias didn’t appear to change a model’s propensity to label text by Black authors as more toxic than white authors. In the interest of thoroughness, the researchers embarked on a proof-of-concept study involving relabeling examples of supposedly toxic text whose translations from AAE to ‘white-aligned English’ were deemed nontoxic. They used OpenAI’s GPT-3 to perform the translations and create a synthetic dataset — a dataset, they say, that resulted in a model less prone to dialectal and racial biases.”

The researchers acknowledge that re-writing Black users’ posts to sound more white is not a viable solution. The real fix would be to expose AI systems to a wider variety of dialects in the original training phase, but will developers take the trouble? Like many people, once hate-speech detection bots become prejudiced it is nigh impossible to train them out of it.

Cynthia Murrell, April 5, 2021

Google and Cookies: Crafting Quite Tasty Bait

March 19, 2021

I read “Alphabet: Five Things We Know about Google’s Ad Changes after Cookies.” I approached the write up with some interest. Cookies have been around for a long time. The reason? They allowed a number of interesting functions, including tracking, cross correlation of user actions, and a covert existence.

Now, no more Google cookies.

The write up explains what Google wants keen observers, real journalists, and thumbtypers to know; to wit:

  1. Privacy is really, really important to Google—now. Therefore, the GOOG won’t support third party cookies. Oh, shucks, what about cross site tracking? Yeah, what about it?
  2. Individuals can be targeted. Those with a rifle shot orientation have to provide data to the Google and use the Google software system called “customer match.” Yeah, ad narrowcasting lives.
  3. Google will draw some boundaries about its data leveraging for advertisers. But what about “publishers”? Hey, Google has some special rules. Yeah, a permeable membrane for certain folks.
  4. FLOC makes non-personalized ad targeting possible. I want to write, “You’ve been FLOC’ed” but I shall not. Yeah, FLOC. But you can always try FLEDGE. So “You’ve been FLEDGED” is a possibility.

How’s this work? The write up does not shed any light. Here’s a question for a “real news” outfit to tackle:

How many data points does a disambiguation system require to identify a name, location, and other personal details of a single individual?

Give up. Better not. Oh, the bait, pivoted cookies. Great for catching prospects I think.

Stephen E Arnold, March 19, 2021

Watching the Future of Talend

March 15, 2021

I read “Talend Sells to Private Equity Firm Thoma Bravo in $2.4 Billion Deal.” I find this interesting. Talend is a software company providing extract, transform, and load services and analytics. Data remain the problem for many thumbtypers fresh from Amazon or Google certification classes. The idea is to suck in legally data from different sources. These data are often in odd ball formats to malformed because another digital mechanic missed a bolt or added a bit of finery. Some people love MarkLogic innovations in XML; others, not so enamored of the tweaks.

What’s Thoma Bravo bring to the table for a publicly traded company with a number of competitors?

I can think of two benefits:

The first is MBA think. Thoma Bravo is skilled in the methods for making a company more efficient. It is a good idea to internalize the definition of “efficiency” as the word is used at McKinsey & Co.

The second is acquisition think. From my point of view, the idea is to identify interesting companies which provide additional functionality around the core Talend business. Then Thoma Bravo assists the Talend management to bring these companies into the mothership, train sales professionals, and close deals.

No problem exists with this game plan. One can identify some indicators to monitor; for example:

  • Executive turnover
  • Realigning expenditures; possibly taking money from security and allocating the funds to sales and marketing
  • Targeting specific market segments with special bundles of enhanced Talend software and business methods.

For more information about Talend as it exists in March 2021, navigate to this link.

Oh, one final comment. Thoma Bravo was involved in making SolarWinds the business success it became.

Stephen E Arnold, March 15, 2021

About TikTok and Privacy: $92 Million Catch Your Attention

March 4, 2021

I have commented about the superficial understanding shared among some “real” and big time journalists of data collection. What’s the big deal about TikTok? Who cares what kids are doing? Dismissive attitude flipped off these questions because “real” news knows what’s up?

ByteDance Agrees to US$92 Million Privacy Settlement with US TikTok Users” suggests that ignoring the China-linked TikTok may warrant some scrutiny. The story reports:

The lawsuits claimed the TikTok app “infiltrates its users’ devices and extracts a broad array of private data including biometric data and content that defendants use to track and profile TikTok users for the purpose of, among other things, ad targeting and profit.” The settlement was reached after “an expert-led inside look at TikTok’s source code” and extensive mediation efforts, according to the motion seeking approval of the settlement.

My view is that tracking a user via a range of methods can create a digital fingerprint of a TikTok user. That fingerprint can be matched or cross correlated with other data available to a specialist; for example, information obtained from Oracle. The result is that a user could be identified and tracked across time.

Yep, today’s young person is tomorrow’s thumbtyper in one of the outfits compromised by the SolarWinds’ misstep. What if the TikTok data make it possible to put pressure on a user? What if the user releases access information or other high value data?

TikTok, TikTok, the clock may be ticketing quietly away.

Stephen E Arnold, March 4, 2021

Business Intelligence, Expectations, and Data Fog

February 10, 2021

Business intelligence and government intelligence software promises real time data intake, analysis and sense making, and outputs with a mouse click. Have you heard the phrase, “I have the Brooklyn Bridge for sale”? Sure, sure, I know I don’t want to own the Brooklyn Bridge, but that super spiffy intelligence software (what I call intelware), count me in.

The marketing pitch for business intelligence and general intelligence software has not changed significantly over the years. In my experience, a couple of nifty outputs like a relationship diagram and a series of buttons set up to spit out “actionable intelligence” often close the deal. The users of the software usually discover three points not making up a large part of the demos, the discussions, and the final contract for the customer’s requirements.

I read “The Age Of Continuous Business Intelligence.” The idea is appealing. Lots of information and no time to read, review, digest, analyze, and discuss the available information. In my opinion, the attitude now is “I don’t have time.”

Yep, time.

The write up asserts:

we [an outfit called KX] know that shortening the time it takes to ingest, store, process, and analyze historic and real-time data is a game changer for businesses in all sectors. Our customers in finance, manufacturing, automotive, telecommunications and utilities tell us that when processes and systems are continuously fed by real-time data that is enriched by the context of historic data, they can automate critical business decisions resulting in significant operational and commercial benefits.

The write up contains a diagram which lays bare “continuous business intelligence.”


The write up concludes:

As the research clearly shows, real-time data analytics is a critical area of investment for many firms. To ensure maximum value is derived from these investments, it is imperative that organizations – regardless of size and sector – challenge their understanding of what real-time means. By implementing a strategy of continuous business intelligence, firms can dramatically reduce the time it takes to uncover and act on insights that can materially change the game in terms of growth, efficiency and profitability.

I love that “research clearly shows.” The challenges for the continuous thing include:

  • Defining real time. (According to research my team did for a project years ago, there are numerous definitions of real time, and there is a Grand Canyon sized gap among these.)
  • Making clear the computational short cuts necessary to process “fire hoses”. (Yep, these compromises have a significant impact on costs, validity of system outputs, and the mechanisms for issuing meaningful outputs from sense making.)
  • Managing the costs. (Normalizing, verifying, processing, storing, and moving data require human and machine resources. Right, those things.)

Net net: Software whether for business or government applications in intelligence work only if the focus is narrow and the expectations of a wild and crazy MBA are kept within a reality corral. Otherwise, business intelligence will shoot blanks, not silver bullets.

Oh, KX is hooked up with a mid tier consulting firm. What’s that mean? A sudden fog has rolled in, and it is an expensive fog.

Stephen E Arnold, February 10, 2021

Dashboards Evil? Worth a Thought. Nah, Just Take What Is Output

February 1, 2021

Business intelligence (BI) is jargon for the technologies and strategies used to manage business data analytics. It is a fancy term for standard operating procedures and looks good on a resume, but one IT CEO wants to make it obsolete. Diginomica discusses how BI could head to the recycling bin in the article: “ThoughtSpot CEO-‘I Want To Kill BI And I Want All Dashboards To Die.’”

COVID-19 has changed global business practices and technology experts spent 2020 investigating ways to aggregate business data. ThoughtSpot CEO Sudheesh Nair explained in the article that traditional aggregation patterns do not apply anymore and companies need to change in order to maintain their customers. Nair believes his ThoughtSpot platform, described as ‘Google for numbers,’ will deliver key insights the same way Google provides information.

Nair pointed out that opinions are easily accessible via a Google search, but facts are harder to find in the endless search results. Nair wants his ThoughSpot platform to make facts as easily accessible as opinions. ThoughtSpot combines hardened facts with a NLP interface to make finding facts easier, think Windows 95 versus the old command land interface:

“ThoughtSpot does this by allowing users to search enterprise data with hyper-personalized questions using natural language processing. It aims to not only give a result for the question you ask, but then also uses AI to offer up alternative questions and results that may be helpful. This is very different to traditional BI, which typically offers you a template for which to present historical, aggregate data.”

In other words, Nair wants to sift the information noise from facts. Today’s BI dashboards offer a plethora of information, but lack personalization notes that could win new customers and retain older ones.

ThoughtSpot will supposedly combine old data with new data to push out BI dashboards and create a new data analytics space for businesses. Nair’s description of ThoughtSpot is an interesting pitch, but it sounds more like a new way to search information. Instead of explaining how ThoughtSpot works it would be better to offer demonstrations of its capabilities.

But thinking? Not the core competency of the thumb typing generations.

Whitney Grace, February 1, 2021

Twitter and the Fire Hose for Academics

January 29, 2021

I read “Enabling the Future of Academic Research with the Twitter API.” According to the official Twitter statement:

Our developer platform hasn’t always made it easy for researchers to access the data they need, and many have had to rely on their own resourcefulness to find the right information.

Understatement, of course.

The post continues:

We’ve also made improvements to help academic researchers use Twitter data to advance their disciplines, answer urgent questions during crises, and even help us improve Twitter.

Help is sometimes — well — helpful. But self help is often a positive step; for example, verifying the actual identity of a person who uses the tweeter thing. There are some software robots chugging along I believe.

Also, charging a subscription fee. The amount is probably less important than obtaining verifiable bank information. Sure, some software robots have accounts at outstanding institutions like Credit Suisse and HSBC, but whatever account data are available might be helpful under certain circumstances.

But academics? How many academics work for non governmental or governmental entities as experts, analysts, and advisors? Will the tweeter thing’s new initiative take such affiliations into account before and during usage of Twitter data?

I assume that a tweeter senior manager will offer an oracular comment like, “For sure.”

There are three hoops through which the agile academic must jump, and I quote:

  1. You are either a master’s student, doctoral candidate, post-doc, faculty, or research-focused employee at an academic institution or university.
  2. You have a clearly defined research objective, and you have specific plans for how you intend to use, analyze, and share Twitter data from your research…
  3. You will use this product track for non-commercial purposes….

Sounds like a plan which will make some nation states’ academics wriggle with anticipative joy.

My view is that this new initiative may unfold in interesting ways. But I am sure the high school science club managers have considered such possibilities. Why who would hire a graduate student to access tweeter outputs to obtain actionable information for use by a country’s intelligence professionals? The answer in the twitterverse is, “Who would risk losing the trust of Twitter by doing that?” Certainly not an academic funded by an intelligence or law enforcement entity.

Right, no one. Misuse the tweeter? Inconceivable.

Stephen E Arnold, January 29, 2021

Tweet This! Real News Discovers the Concept of Hidden in Plain Sight

December 31, 2020

Remember the Purloined Letter? No, that’s okay. Thumbtypers don’t either. I read “Just How Bad Was This Year? These Professors Found Answers on Twitter.” I noted this passage:

Since 2008, the duo [professors at a school in Vermont] has taken a random 10 percent of everything tweeted each day, seeking truths hidden in plain sight. (Whileacknowledging, as Danforth put it, that “Twitter is a nonuniform subsample of utterances made by a nonuniform subsample of humans whoare on the Internet.”) They’ve used it, for example, to explore fame, finding that DonaldTrump and K-pop band BTS are mentioned as commonly as some regular words (think: “after,” “would.”). As Dodds put it, “The word‘Trump’ has been in the top 300 words all year this year, which he’s never done before. That’s more common than the word ‘God.’ ”

The sampling is done by the Hedonometer, possible a reference to either a town in England or a unit of pleasure used to theoretically weigh people’s happiness. I like the latter candidate, split infinitive, and the weird idea of “weighing” happiness. I often say to the grocery clerk in Harrod’s Creek, Kentucky, “I will take a pound of happiness and a half pound of ricotta, please.”

The big find seems to be:

Some trends have emerged through the years. All else being equal,Saturday is the week’s happiest day on Twitter, Tuesday the saddest.National holidays cause huge spikes in happiness, with Christmas beingthe most cheerful. Major sporting events and birthdays of pop stars,particularly K-pop stars, tend to make for gleeful days. On the flipside, natural disasters and mass shootings tend to spark more unhappydays.

What’s the analysis reveal?

“In the last five years, we’ve seen the usual weekly cycle justget busted,” Dodds added. “It’s sort of all over the place now.Events are happening any day of the week. It’s much more what Iwould call emotional turbulence.”

Remarkable in a way, a modest way.

Stephen E Arnold, December 31, 2020

Saddle Up, Statistical Analytics Fans. Place Your Bets on AI

December 7, 2020

The Best Way to Win a Horse Race? Mathematicians May Have the Answer” is an interesting example of a snappy headline not supported by the write up’s text. For gamblers, the promise of finding a way to predict which mighty steed will cross the finish line first is catnip. (Sorry for the mixed animal metaphor, but I could not resist.) The problem is that the summary of the study includes lots of references to data collection and number crunching. Then the killer statement:

Various other scientific attempts to explain performance over the past 4 decades “haven’t been particularly successful,” he says—and not just because horses vary so much in body size and aerobic capacity: The models cannot account for the horse’s own behaviors. For example, a horse might give up when another horse passes it, because it doesn’t understand that it’s supposed to win. Until researchers can get inside the horse’s head and account for psychological variables, Knight says, “we can’t truly model performance.”

Net net: If one can’t model an equine, what’s that suggest for figuring out what human will cross an innovation finish line, crack a tough problem, or write a headline which does mislead the pony player?

Stephen E Arnold, December 7, 2020

Clarity: A Better Name Than Pluton. Pluton?

November 20, 2020

After two years, Clarity has finally made it out of Beta, we learn from “Microsoft Clarity Debuts as Free Analytics Tool with Heat Maps” at Search & Performance Marketing Daily. The free tool uses heat maps to analyze the behavior of visitors to one’s website. Reporter Laurie Sullivan writes:

“Clarity — designed to have a low impact on page-load times and there are no caps on traffic no matter what the number of visitors to the website — helps give marketers a deeper understanding of why at website performs one way and not another. It also provides anonymized heat maps and data that show where site visitors clicked and scrolled, and enables marketers to analyze use behavior on the website exactly as it happened through a job description code. Some of the data includes the name of the browser, and whether they are using a PC, tablet or mobile phone to access the site. Heat maps provide a visual way to examine large numbers of site visitor interactions. Microsoft built two types: click maps and scroll maps. While the heat maps tell marketers which pages get the most clicks, the click maps tell marketers what website page content visitors interact with the most. Areas in the map marked in red have the highest frequency of clicks and are usually centered on focal points.”

The heat maps let marketers know whether visitors are clicking where they want them to. It also reports certain behaviors—excessive scrolling, dead clicks, and rage clicks. The last term describes users clicking several times on a spot they believe should be a hyperlink but is not—one would want to either fix an intended link or tweak the graphics on those spots. The tool also supplies a dashboard that presents metrics of the overall traffic patterns, time spent on the site, and even concurrent JavaScript errors. Microsoft pledges Clarity complies with the EU’s General Data Protection Regulation.

But Pluton, Microsoft’s mystery processor? Pluton?

Cynthia Murrell, November 20, 2020

Next Page »

  • Archives

  • Recent Posts

  • Meta