Mediaboss Marketing

Remember tags? Around2007, I was all overthem, and I feel no shameabout that. Well, notmuch shame. [Davidsbook, Everything Is Miscellaneous,was published in 2007. Ed.] Lettingusers apply whatever tags, or folksonomies,they wanted to digital contentblew apart constraints on knowledgethat we'd assumed for millennia were strengths of knowledge. In fact, the idea that each thing had only one real tag was the bedrock of knowledgefor thousands of years: A tomato is a vegetable, not some other thing.

Ok, nerds, youreright; a tomato is actuallya berry. But youre justproving my point: We liketo think that a thing is one thing and notany another. At least in some contexts.

Of course, before tags, we would applymultiple classifications to things: A bookabout tomatoes might get classified underrecipes, healthy foods, and the genusSolanum. But a tomato is also a classicallyred object, roundish, delicious, squishy, asource of juice, a bad thing to learn jugglingwith, something we used to throwat bad actors and corrupt politicians, andso much more.

Then, with sites that allowed userbasedtagging, users could tag tomatoeswith whatever attributes were importantto the user at that time. We can now dothis with the photos we take, the placeswe go on our maps, the applications weuse, the sites we visit, the music we listento. Tags have become so commonthat theyve faded from consciousnesssince 2007, although sometimes a cleverhashtag pops up.

While AI in the form of machine learningcan automatically apply tags, it mayreduce the need for tags. Already we cansearch for photos based on their content,colors, or even their mood and all withoutanyone attaching tags to them.

Machine learning redefinestagging

But more may be at stake. Mightmachine learning complete the conceptualjob that tagging began, leading us from adefinitional understanding of what thingsare to a highly relational view? My prediction(My motto: Someday Ill get oneright!) is that within the next few years,dimensionality is going to become animportant, everyday word.

One view of meaning is that a wordis what its definition saysit is, as if a definition werethe long way of saying whatthe word says more compactly.But thats not howwe use or hear words. InThe Empire Strikes Back,when Princess Leia says, Ilove you to Han Solo andhe replies, I know, thedefinitions of those wordscompletely miss what justtranspired.

Tagging has made clear that thingshave very different meanings in differentcontexts and to differentpeople. Definitions havetheir uses, but the timeswhen you need a dictionaryare the exception. Tags makeexplicit that what a thing is(or means) is dependent oncontext and intention.

Machine learning is gettingus further accustomedto this idea, and not just for words. Forexample, a medical diagnostic machinelearning model may have been trained onhealth records that have a wide variety ofdata in them, such as a patient's heart rate and blood pressure, weight, age, cholesterollevel, medicines theyre taking, past history, location, diet, and so forth. The more factors, the more dimensions.

Link:
Tags, AI, and dimensions - KMWorld Magazine