In­no­va­tion, or How Thom­son Reuters Won the Other In­for­ma­tion War

Policy - - In This Issue - Shawn Mal­ho­tra

The brand that is now fre­quently cited as Canada’s most glob­ally rec­og­nized was born in the min­ing towns of North­ern On­tario in the early 1930s. Roy Thom­son, the son of a Toronto bar­ber, launched North Bay’s first ra­dio sta­tion, then the Tim­mins Daily Press. To­day, Thom­son Reuters is the world’s largest mul­ti­me­dia news agency, pro­vid­ing busi­ness and in­ter­na­tional news, fi­nan­cial mar­ket data and reg­u­la­tory in­for­ma­tion to the fi­nan­cial, le­gal, tax and ac­count­ing in­dus­tries as well as a range of other con­tent to con­sumers around the world.

Thom­son Reuters has been serv­ing pro­fes­sional knowl­edge work­ers in more than 100 coun­tries for more than 100 years. As all good Cana­di­ans do, we bring a grounded, com­mon-sense ap­proach to the way we run our global busi­ness. But, per­haps the most in­her­ently Cana­dian trait we share with all en­dur­ing busi­nesses is a healthy fo­cus on in­no­va­tion.

We did ‘Big Data’ long be­fore the term be­came cor­po­rate jar­gon. We be­gan ap­ply­ing ma­chine learn­ing 15 years be­fore Watson played Jeop­ardy. To­day, our AI tools use so­cial me­dia

to ver­ify fact from fic­tion and en­able our jour­nal­ists to fil­ter out so­cial me­dia noise and iden­tify break­ing news events.

Ours is a great story, and a very Cana­dian one. Thom­son be­gan as a news­pa­per com­pany in North­ern On­tario in the 1930s. Over the decades, the busi­ness ex­panded across in­dus­tries and bor­ders, in­clud­ing a highly suc­cess­ful own­er­ship stake in a North Sea oil and gas con­sor­tium. But the com­pany’s lead­ers had the fore­sight to know that oil was a non­re­new­able re­source. They needed to find the oil wells of the 21st cen­tury: in­for­ma­tion.

Fast-for­ward to roughly 30 years ago—Thom­son was a hold­ing com­pany that pub­lished about 200 news­pa­pers, along with text­books and pro­fes­sional jour­nals, as well as the largest leisure travel busi­ness in the UK. Thom­son had the fore­sight to un­der­stand the sweep­ing change the in­ter­net and dig­i­ti­za­tion of con­tent would bring and started to di­vest its print as­sets for higher-mar­gin dig­i­tal, pro­fes­sional in­for­ma­tion and ser­vices as­sets.

While at its core, Thom­son re­mains an in­for­ma­tion pub­lish­ing com­pany, early in­vest­ment in elec­tronic de­liv­ery be­came a cor­po­rate pri­or­ity. At the time, the Thom­son Cor­po­ra­tion pro­vided much of the spe­cial­ized in­for­ma­tion con­tent the world’s fi­nan­cial, le­gal and re­search or­ga­ni­za­tions re­lied on to make busi­ness-crit­i­cal de­ci­sions. In 2008, the com­pany bought Reuters Group, a global fi­nan­cial in­for­ma­tion and news busi­ness.

Through­out our jour­ney, we have been in­no­vat­ing and build­ing a com­pany de­signed to com­pete in the In­for­ma­tion Age. Cus­tomers used to pay for printed vol­umes of need-to-know data. They moved on to net­works of in­for­ma­tion stored in data­bases and de­liv­ered in elec­tronic form. To­day, they pay for the right an­swer, de­liv­ered at the right time when they need it in their work­ing lives. With the ex­plo­sion of data and pro­lif­er­a­tion of “free” in­for­ma­tion, it has be­come harder than ever to ex­tract true value from this wealth of op­por­tu­nity. The foun­da­tion of our in­no­va­tion ef­forts has been the work we do un­der the hood. Need-to-know pro­pri­etary data is im­por­tant. But, the key has al­ways been the in­for­ma­tion ar­chi­tec­ture around the data that en­ables us to ex­tract value-added mean­ing from the in­for­ma­tion.

To make bet­ter use of the data we had, it needed to be “freed” from the si­los it was cre­ated for and man­aged in. So, in the early 1990s, Thom­son Reuters be­gan phas­ing in ar­ti­fi­cial in­tel­li­gence, nat­u­ral lan­guage pro­cess­ing and ma­chine-learn­ing tech­nolo­gies to with in­creas­ing so­phis­ti­ca­tion. Back in 1975, the CN Tower had just been com­pleted and jour­nal­ists were still writ­ing sto­ries on type­writ­ers and fil­ing them by phone. There was no in­ter­net and no per­sonal com­put­ers. West Pub­lish­ing (a fu­ture di­vi­sion of Thom­son) launched West­law, one of the first on­line le­gal re­search ser­vices. At­tor­neys used ‘dumb’ ter­mi­nals to dial-up to a main­frame. The con­tent was lim­ited (disk space was ex­pen­sive), the search lan­guage sim­plis­tic. Soon the search was en­hanced to al­low the use of Boolean terms. Full text search only came much later. Over the next decade, the con­tent ex­panded sig­nif­i­cantly, but search en­gine tech­nol­ogy re­mained much the same.

In 1992, we launched the first com­mer­cial nat­u­ral lan­guage search en­gine. It was the first search en­gine in-mar­ket to in­tro­duce prob­a­bilis­tic ranked searches for nat­u­ral lan­guage queries—a form of ma­chine learn­ing. This pro­gram used sta­tis­tics to pro­vide an es­ti­mate of what an­swer is the one the lawyer is prob­a­bly look­ing for. Be­fore this, re­sults were sim­plis­ti­cally or­dered and users had to wade through long lists of ir­rel­e­vant re­sponses to find their an­swer.

We were also one of the ear­li­est in the in­for­ma­tion in­dus­try to in­tro­duce full ma­chine-as­sisted au­to­ma­tion at scale for text-min­ing and con­tent-en­hance­ment tech­nolo­gies. This en­abled the search of mass amounts of un­struc­tured data and dra­mat­i­cally re­duced the time it takes to sift through hun­dreds of le­gal doc­u­ments.

By 2000, the in­ter­net and the web were see­ing ex­po­nen­tial growth. Ma­chine learn­ing ap­proaches for many in­for­ma­tion tasks started to get more and more trac­tion. We cre­ated ma­chine learn­ing tech­nol­ogy that en­abled us to man­age the mas­sive scale of data we had. And, in con­trast to

Thom­son had the fore­sight to un­der­stand the sweep­ing change the in­ter­net and dig­i­ti­za­tion of con­tent would bring and started to di­vest its print as­sets for higher-mar­gin dig­i­tal, pro­fes­sional in­for­ma­tion and ser­vices as­sets.

The foun­da­tion of our in­no­va­tion ef­forts has been the work we do un­der the hood. Need-to-know pro­pri­etary data is im­por­tant. But, the key has al­ways been the in­for­ma­tion ar­chi­tec­ture around the data that en­ables us to ex­tract val­ueadded mean­ing from the in­for­ma­tion.

the gen­eral pub­lic, our cus­tomers (lawyers, ac­coun­tants and bankers) had very spe­cific in­for­ma­tion needs.

In short, while a Google search al­lows users to ask a sim­ple ques­tion and re­ceive a factual com­pi­la­tion of in­for­ma­tion, a West­Law search goes a step fur­ther—the soft­ware comes back to the in­quirer with a com­plete set of ju­rispru­dence. The sweat and hours that lawyers would have de­voted to un­earthing the in­di­vid­ual pieces of in­for­ma­tion needed for un­der­stand­ing a prece­dent are now han­dled by our soft­ware. The plat­form in­cludes nat­u­ral lan­guage search, which fur­ther sim­pli­fies the way le­gal re­search is con­ducted, help­ing re­searchers find an­swers in one tenth of the time.

In the last twenty years, the amount of in­for­ma­tion has ex­ploded—for all busi­ness pro­fes­sion­als. The quan­tity is over­whelm­ing, and it is ac­cel­er­at­ing rapidly. For con­text, more data has been cre­ated in the past two years than in the en­tire pre­vi­ous his­tory of the hu­man race. At Thom­son Reuters, we now process and col­lect more data in a sin­gle day than we did in a month five years ago. Even as we move to the cloud, we still store 60,000 ter­abytes of data in our data cen­ters. To put that in con­text, the U.S. Li­brary of Congress con­tains 200 ter­abytes of data and the to­tal size of Wikipedia is 3 ter­abytes. Thom­son Reuters data is used to price $3 tril­lion in as­sets daily—nearly 2.5 mil­lion price up­dates per sec­ond.

We be­lieve that the key to ex­tract­ing value is to do more with data. In order to ef­fec­tively use data, it’s im­por­tant to un­der­stand how it con­nects to the real world. By us­ing shared plat­forms and work­ing across our busi­nesses we are mak­ing our data more ac­ces­si­ble and valu­able for our cus­tomers, no mat­ter how they ac­cess it. Our cus­tomers rely on us for the an­swers they need.

To­day, across Thom­son Reuters, we use our sub­ject mat­ter ex­perts, ar­ti­fi­cial in­tel­li­gence and ma­chine learn­ing to con­tin­u­ally im­prove how we find, ex­tract, tag and struc­ture data. We fuse world-class con­tent with emerg­ing tech­nol­ogy and deep do­main ex­per­tise en­sur­ing our an­swers stay ahead of the curve. We have trans­formed our­selves from a pub­lish­ing com­pany into an in­for­ma­tion and tech­nol­ogy com­pany. Our cus­tomers are de­pen­dent on know­ing about events and risks that can af­fect their com­pa­nies, their clients, mar­kets or sup­ply chains. Stay­ing up-to-date is crit­i­cal, but the amount of data that is pro­duced daily is over­whelm­ing. We use AI tech­nolo­gies to au­to­mat­i­cally con­sume and an­a­lyze the fire hose of data from news, mar­kets and so­cial me­dia.

Be­hind the scenes, AI tech­nolo­gies have been de­ployed across Thom­son Reuters. The vast data sources that we have cre­ate an al­most un­lim­ited num­ber of op­por­tu­ni­ties for spe­cial­ized in­for­ma­tion ex­trac­tion. The var­i­ous so­lu­tions fur­ther ex­panded our knowl­edge base and con­nected the con­tent, mak­ing re­search eas­ier and en­abling new forms of an­a­lyt­ics.

That is why we are in­vest­ing in tech­nol­ogy. Glob­ally, we in­vest more than $3 bil­lion per year on tech­nol­ogy. We have more than 12,000 soft­ware en­gi­neers, systems ar­chi­tects and data sci­en­tists around the world who de­sign and de­velop prod­ucts that ad­dress the com­plex needs of con­duct­ing busi­ness in to­day’s world and advance our cus­tomers’ ex­pe­ri­ences.

In 2016, we opened our Thom­son Reuters Tech­nol­ogy Cen­tre in down­town Toronto, which is also home to our global Cen­tre for AI and Cog­ni­tive Com­put­ing. The Toronto Tech­nol­ogy Cen­tre is ex­pected to cre­ate 400 new tech­nol­ogy jobs by the end of 2018 and up to 1,500 jobs in to­tal. In the fall of 2017, we an­nounced we are in­vest­ing $100 mil­lion in a per­ma­nent lo­ca­tion for our tech­nol­ogy cen­tre.

We now process and col­lect more data in a sin­gle day than we did in a month five years ago. Even as we move to the cloud, we still store 60,000 ter­abytes of data in our data cen­ters. To put that in con­text, the U.S. Li­brary of Congress con­tains 200 ter­abytes of data and the to­tal size of Wikipedia is 3 ter­abytes.

Ten years after the first iPhone®, a new dig­i­tal world pow­ered by big data, cog­ni­tive com­put­ing and the cloud prom­ises to change the way we live, work and in­ter­act. We have been a pi­o­neer of dig­i­tal prod­uct de­vel­op­ment for decades. From us­ing Blockchain to bring de­vel­op­ing coun­tries the con­fi­dence of se­cure land records to us­ing ma­chine learn­ing to help jour­nal­ists and read­ers alike sep­a­rate fact from fic­tion. We are ap­ply­ing cut­ting-edge tech­nolo­gies to emerg­ing chal­lenges.

Thomp­son Reuters photo

Thom­son Reuters employees are work­ing to em­bed emerg­ing tech­nolo­gies into cus­tomer so­lu­tions as part of a $100 mil­lion in­vest­ment in a new Toronto Tech­nol­ogy Cen­tre that will cre­ate up to 1,500 jobs.

Newspapers in English

Newspapers from Canada

© PressReader. All rights reserved.