analytics continues from page 1 - knowledge management · world september 2013 in a business...

1
This image shows an output generated by Palantir’s system of Haiti earthquake data.

Upload: duongthuan

Post on 03-Jul-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

16 September 2013KMWorld www.kmworld.com

In a business environment where nothingseems to be set in stone, data scientists have beenbrought into the executive suite. Numbers are thenew currency of success. The e-book best sellersinclude study aids for advanced mathematics. Thefact that those books appeal to business executiveswho took only required general math underscoresthe escalation of what John Allen Paulos called“innumeracy.” Analytics is the new black. Statis-tics delivers a competitive advantage. Knowledgeof the dark arts of calculation delivers results in aworld chock full of MBA mumbo jumbo, and mar-keting speak is making a comeback.

The shift One of the most interesting knowledge manage-

ment shifts I have witnessed is the bulldozer-likepush in organizations for analytics: the shift fromMadison Avenue recipes for making sales and theconsulting firm’s epic cheerleading for managementchange, to numbers that are more concrete. Withoutdata, no assertion will get more than a single Pow-erPoint slide. Metrics, charts, graphs and Holly-wood-style animations are needed to captureattention and build support for a business decision.

The problem is that in most organizations thenumber of individuals who have highly refinedmath expertise is usually small. Even in knowl-edge factories operated by the likes of IBM(ibm.com) and Google (google.com), the mathwizards are outnumbered by individuals withsofter skills in law, public relations, sales, mar-keting and business development. Even in theaccounting units, the bean counters work withnumbers within generally well-defined domains.A change in tax law does not require venturinginto the murky world of numerical procedures thatcannot be calculated with today’s computers orresults that apply to n-dimensional probabilisticoutputs. There is a difference between basicapplied math and the sharp edges of multivariateadaptive regression splines.

Not surprisingly, a number of new and quiteinteresting products and services have becomeavailable. Those solutions are different from theold school systems that are often taught inadvanced classes. You may have encountered SAS(sas.com) or SPSS (IBM) in statistics, or MAT-LAB (mathworks.com) or Mathematica (wolfram.com) in a business course or a third-yeargenetics class. But today’s analytics bridge theGrand Canyon between basic numerical methodsand the Star Trek world of number theory.

Some innovatorsTIBCO (tibco.com), whose name is an

acronym for “the information bus company,” is aninnovator in enterprise systems. In 2007, TIBCOpaid about $200 million to acquire Spotfire, acompany founded by Christian Ahlberg. TIBCOwas one of the first of the enterprise solutionsfirms to move aggressively into the advanced ana-lytics market. Spotfire was an important acquisi-tion because it delivered to TIBCO a system andmethod that combined interactive interfaces to awide range of data without having a programmeror specially trained technical intermediary for theexecutive. According to TIBCO, its infrastructure

made data available in near real time, and Spot-fire allowed TIBCO’s licensees to explore and dis-cover relationships, trends and outliers oranomalies with a mouse click or two.

As advanced as Spotfire is, the founder of Spot-fire set up a new company to tackle the challenge ofthe large-scale data flows available from Web-enabledapplications. Christian Ahlberg founded RecordedFuture (recordedfuture.com) and obtained financialsupport from In-Q-Tel (iqt.org), the investment armof the U.S. intelligence community, and Google.Recorded Future tackles “Web intelligence” derivedfrom open source content. For an organization, theRecorded Future system can be used to derive com-petitive intelligence from public information aboutbusinesses and from publicly accessible content suchas patent applications.

Another company that is pushing the bound-aries of advanced mathematics is Agilex(agilex.com). The company supports a number ofU.S. government agencies, but the firm’s tech-nologies are gaining attention in the financial sec-tor. The low-profile firm employs about 400professionals. The firm’s analytics functions areembedded in a range of applications that includehealth information exchanges, data visualizationand discovery.

Agilex says its Phanero system “applies a sophis-ticated mathematical technique to dynamically dis-cover relationships and key concepts in large volumesof structured, semi-structured and unstructured data.At the same time, Phanero’s integrated Entity Extrac-tor automatically identifies and tags entities of inter-est, such as persons, locations or organizations withinthe data. This makes it possible to analyze, map, rankand explore complex associations between relevanttopics and specific objects.”

For the organization licensing the system, Agilexreports, “Phanero addresses a variety of advancedanalytics and discovery requirements. PhaneroInvestigations is a Web-based user interface optimizedfor the unique requirements of the intelligence and lawenforcement communities. For example, it provides

automated risk assessments based on strength of rela-tionship. It also integrates directly and automaticallypopulates a number of custom and commercial datavisualization and case management tools, such asPalantir (palantir.com) Government.”

A slower revolutionThose three examples make clear a progression

in enterprise analytics. TIBCO was an early leaderin integrating advanced analytics within a broadermessaging capability of its bus architecture. Thefounder of Spotfire shifted his focus from trimmedor constrained data sets to the rapidly growing andunbounded environment of public Web content.Agilex combined management tools, applicationsand mathematical methods into integrated solu-tions that support other specialist vendors’ toolssuch as the Palantir discovery application.

The progress in advanced analytics, however,is not moving at the pace of the mobile handsetsector or social networking. The mathematics rev-olution is underway but it works at a slower pace.One of the friction points in advanced analytics isthe computational capabilities of today’s computersystems. Most of the individuals with training inadvanced mathematics know about the Big Oproblem. The snappy name means that some typesof numerical recipes cannot be calculated giventhe time and resource constraints in an organiza-tion. If you are technically savvy, you know thatprogrammers have to dodge the class of decisionproblems known as NP-complete, where there areno known fast solutions. Too ambitious recipescan leave the computer huffing and puffing with-out finishing the calculation.

A second point of friction for analytics systems is thegraphical interface. The math used in most of the sys-tems, including open source solutions like Talend(http://sourceforge.net/projects/talend-studio) is sec-ondary to the interface and presentation layer. A personwith math expertise can use most systems after a shortperiod of familiarization. However, a product managerrequires an interface that works more like an automo-

ANALYTICS continues from page 1

This image shows an output generated by Palantir’s system of Haiti earthquake data.

KMW_9_2013_ 8/8/13 1:47 PM Page 16