War for Talent: Realigning Academic Pursuits with What Business Demands

Shekhar Murthy is presenting this paper on behalf of herself and Padmaja N

Is war for talent for real?

Geography? In the changing globalized and competitive world, war for talent is escalating. Excess supply of talent in places like Mexico, Brasil, India.

Sociology? Who is affected? IT, Manufacturing, Marketing, Finance

Emerging tech advancements across the globe:

• Mexico ? construction
• Poland ? high tech, automotive
• Russia ? software engineering
• S Korea ? ICT, Genomics
• China ? Telecom, petrochem, pharma biotech nanotech
• Taiwan ? semiconductors

IT manpower gap (2009) Alarming shortfall in manpower. If you look at the IT scenario in India the total demand is 1,120,000 but total supply is only 885,000

Decreasing workforce: future recruitment crisis, demand for offshore talent, employable talent, skill gap, developing talent, hiring and retention of human capital, balance of power (employees have the power because of high demand), talent management

Emerging markets: countries in the transitional phase between the developing and developed status

Global talent supply: Asia produces 45% of science and tech engineering graduates Future growth of industry dependent on this talent

The way ahead:

• Sourcing People: Influence the quality of resource base, focus on enhancing supply
• Developing People: Continually move people from the learner category to expert category
• Engaging People: Talent management, nurture professional aspirations

Because of this competition for world talent, instead of looking for existing talent, it’s time to generate the talent

Focusing on the supply base: nurturing talent with 21st century skills: adaptability, high order thinking skills, sound reasoning skills with quality consciousness, lifelong self-directed learning, interactive behavioral and social competencies, result oriented professionals with practical knowledge
Pedagogy as the enabling factor: evaluation parameters

Portability, flexibility, scalability, economics of investment, collaborative learning, asynchronous mode of learning

Case study: MSIT creating global IT professionals

Their program is entirely focused on preparing students for their professional careers.

• Learning by doing, problem/story centered curriculum: Create virtual environments with role play — students taking the part of VP Engineering or VP Sales
• Soft Skills
• Team-based learning, collaborative work
• Mastery model — instead of 40 or 50% they have to get between 70-90% to pass
• Mentors ? for every 10 students they have one mentor
• e-Learning ? rich resources ? course being offered in collaboration with Carnegie Mellon University

Salient findings of case study

• Project-centric curriculum has been a critical success factor
• Soft skills training has turned students into all rounded professionals
• Instructor led component to be strengthened
• Internships can give the professional edge
• Weblogs and focus learning groups can promote knowledge sharing

Conclusion
Good relationships between academia and industry
Learning emphasizing on practical knowledge
Mentoring with collaborative learning
Introduction of mobile learning, podcasting and game-based learning

Swarm Shopping: The Simulation of Herd Behavior

Supermarkets have two opposite goals: retailers want to keep the customers in the store as long as possible, customers want to leave as soon as they’re done shopping

More profit and sales are the number one priority for every manufacturer to live, succeed and stay in business

Two ways:

1. out?store tactics (direct mail, etc.)
2. in-store tactics (bargains, everyday low prices, limited offers, etc.)

They looked at real-time datamining, i.e., while customers are in the market

Impulse shopping: pure impulse, reminder impulse, suggestion impulse, planned impulse (sale or bargain)
Impulse shopping accounts for 40% of all shopping in supermarkets (wow!)

He keeps skipping past interesting-looking slides

People have a natural tendency to follow the crowd ? knowing what is being purchased by other people in real-time will affect others to purchase the same product

Ideally the implementation of our model can provide customers with a better shopping experience and retailers with a higher sales level, thus easing the tension that we described earlier between the customers’ and the stores’ goals

Using RFID shopping carts, provide users with real-time information about popularity of items, real-time, personalized offers based on what they’re currently buying

The Swarm-Moves simulator attempts to capture the collective average choice that customers are taking ? no need to worry about WHY (age, sex, status, pay, etc.), just WHAT

Database contains the inventory of swarm mall (the virtual supermarket)

120 products with various attributes

Customers do not care about others’ purchases until it reaches a certain threshold

Goes through the simulation and the equation that produces suggestions, based on that threshold

Swarm model is able to increase sales by 29% over just using discounts and promotions

How will people trust the system? The supermarket can just add ‘cheaters’ to influence the count. So they tested that, and found that they would need as many cheaters as customers in order to influence the system! Not feasible (KC note: it’s really a silly argument. No customer will ever investigate the technology to find out that level of detail — they’ll either trust the system or they won’t. And in the end it may not even matter — everybody knows laugh tracks are fake but we still laugh louder and longer if they’re used.)

Effective marketing tool for supermarkets to increase their sales volume based on collective choice of customers

Environmental information can be used to influence customers to buy more on impulse

Future work: Addition of multiple entrance and exits, combine bargains, effect of other factors, like price, weight, size, etc.

With RFID, amount due is already prepared when you get to the front of the store

KC: I think this guy could be a billionaire in the next year or two…

GroupShare — Designing an Awareness-Rich Online Groupware System

Groupware
Groupware is software to support groupwork
It can include online learning environments, email, discussion forums
Groupware can be organization-based (e.g., Lotus Notes, MS Exchange), or fully online (e.g., BSCW, Yahoo! Groups)
Online groupware is becoming increasingly sophisticated, and is often free/cheap to use

Awareness
Awareness defined as ‘an understanding of the activities of others, which provides a context for your own activity’ (Dourish & Bellotti, 1992)
Implicit and taken for granted in face-to-face group work (if all of us fall asleep, he can just see it and stop talking), but all but gone in online group work
Seen as crucial to effective online group work
Awareness mechanisms are features in groupware which help to increase awareness
Awareness systems have to be autonomous ? you can’t ask users to upload their awareness information

GroupShare
GroupShare is a fully online groupware application built by Baatard as basis for ongoing awareness-related research

Built with two core design goals:

• To be as centralized, flexible and generically applicable as possible
• To provide a rich set of awareness mechanisms

Currently implemented in a university environment

Showed demonstration of app ? files (can see comments, open, etc), asynchronous chat, awareness feature, forum, live chat, personal workspace showing files you’ve uploaded and trophies you’ve collected

GroupShare Awareness Mechanisms
GroupShare strives to provide a rich set of awareness mechanisms
These mechanisms recognize both direct and indirect participation to provide a better understanding of group members’ actions
Some mechanisms aim to specifically encourage sustained participation, and provide a better understanding of actions over time

• GroupShare Recent Activity: Simple event-based awareness mechanism, can be customized to suit needs (KC: kind of like FB NewsFeed)
• GroupShare File Statistics: lists number and timeframe of views, downloads and comments on shared files, per user; paints a picture of how group members have responded to a file shared within the group
• GroupShare Trophies: Trophies are collectible rewards for participation-related events and milestones; encourage participation and gives group members an ida of the participation of others (e.g., contributor, silver = 5 submissions). He says it’s a silly thing but people go crazy for it!
• GroupShare Participation Awareness: processes and aggregates logs of all actions to provide an at-a-glance measure of participation

(KC) This is a great system! The awareness features are sorely lacking in BlackBoard, which is the de facto standard for e-learning

Groupware design recommendations

1. Keep the system simple by avoiding superfluous or overly complex features, while still supporting a wide range of group work activities
2. Provide multiple forms of communication, both synchronous and asynchronous
3. Provide a rich set of awareness mechanisms to encourage and support sustained participation and collaboration
4. Minimise the cognitive load of the system by using a simple web-based interface and allowing users to create content in the applications they are accustomed to
5. Avoid building unnecessary amounts of structure or workflow control into the system ? make the system as generic and flexible as possible

Unfortunately, only available for research purposes ? maybe if somebody gives him money for it he’ll make it available elsewhere!

Dr. Drapeau is presenting this paper on behalf of himself, Peyton C. Hurley, and Robert E. Armstrong

Shock and Awe: The Ecology of Counterinsurgency

Paper is a combination of biology and military problem

He’s at a small thinktank at the National Defense University, has recently been heading up a project called ’social software for security’ (how the DoD can use Facebook for defense)

Bio-inspired network science

What is an insurgency? “?a struggle for control over a contested political space, between a state? and one or more popularly based, non-state challengers.” ? David Kilcullen, “Three Pillars of Counterinsurgency”

From Kilkullen, counterinsurgency redux: Pre-existing social networks, self-synchronizing swarm of cells, etc.

Hunter-prey models and COIN? (COIN is just an acronym for counterinsurgency, in case you’re non-military like me)

Pyramid model: Divided people (and how to deal with them) into pyramid: gen population at the bottom (politics), criminals in the middle (policing), insurgents at the top (military)

Football field model: Rebellion (one end zone), population (field), authority (other end zone)

The mathematics of interference competition ? competition in which an individual or population behaves in a way that reduces the exploitive efficiency of another individual or population

SHOCK: the five key behaviors ? search, handle, oppose, combat, kill

AWE: the overall research project ? analogies with ecology

Caveats and comments: measurable variables? Good data? Quantifying political control? Data insurgency-specific or general? Access versus support of population? What outputs will it yield? Descriptive? Prescriptive?

Conclusions

• Ecosystems, like other complex adaptive systems, are applicable to national security
• There are parallels between ecological competition and counterinsurgency warfare
• Using ecology, it may be possible to model COIN in descriptive and predictive ways

Major Ring is presenting this paper on behalf of himself, Stephen Henderson, and Ian McCulloh (the last guy)

Gathering and Studying Email Traffic to Understand Social Networks

Determined that you can ascertain network information just by looking at who sent email to whom

Ike Net Study
Multi-disciplinary, social network analysis

Overall research goal: ascertain underlying social network based on frequency of communication among members within a selected group of participants

Email gathering techniques decentralized data collection (client), centralized data collection (server)

Officers in the USMA Eisenhower Leadership Development program will have their email monitored to develop a database of daily email traffic. This database will be used to develop dynamic social networks. Theories on social and cognitive causes for links in a social network will be studied and verified. New algorithms for detecting changes in networks will also be developed by statistical process methods

Ongoing research, part of a 5-yr study

Email data: email header data only, anonymized data

Technical Background
SMTP: clients compose & read email, server acts on behalf of client to send & receive all email

Information gathered: TO and FROM, Date, subject, message ID
Not concerned with email content

Client Side Data Collection
Developed in Visual Basic, patch installed on user’s computer, advantages (works in absence of common server), disadvantages (unreliable, requires additional processing), may be a necessity (e.g., forensics)

Server Side Data Collection
Developed in Java (could also use Python, Perl), parses MS Exchange Log Files, collects, filters and consolidates data, advantages (more reliable, more data), disadvantages (privacy concerns, data files can be massive)

Takes email, consolidates into usable information, converts to visual graph

Dyad Count
Communication between two nodes, defines three types of communication (mutual, asymmetric, none), total possible combinations: D = N(N-1)/2 (KC: Hey, we learned that in PMBOK training!)

Results
In second week, increased number of asymmetric and mutual communications, coincided with heavy academic requirements & social activities
In fourth week, limited communication coincides with Thanksgiving break

In other words, they could see variations in amounts of certain types of email activity when something was going on (or not going on).

Future Work
Data mining for further analysis, expanding type of collected data (CC & BCC, forwarded messages, replies, encrypted messages), graphical visualization

Major McCulloh is presenting the paper on behalf of himself, Eric Daimler, and Kathleen M. Carley.

Using Term-Frequency-Inverse-Document-Frequency of Email to Detect Change in Social Groups

TF-IDF identifies terms that are uniquely important to a document (discarding words like ‘is’, ‘the’, ‘a’; keeping words that are frequent in a particular document but not in a large body of documents) ? provides a series of character strings that are meaningful

Looks at those character strings and at the cosine similarity between weeks.

Their project applied to a quarter million emails from the Enron corpus and took subset of internal Enron emails only, aggregated it by weeks over 138 weeks. Ended up with TF-IDF vector that describes the sentiment for what the Enron Corporation was talking about in that week.

The cosine similarity between TF-IDF vectors and a reference vector provide a stochastic sequence of temporal numbers ? shows formula for average reference vector, then takes each weekly vector and takes the cosine similarity against the average vector to get change over time.

Statistical process control

• Used in manufacturing to maintain quality control
• Monitors a process for potential change
• Calculates a statistic from observed measurements of a process and compares it to a decision interval
• If the statistic exceeds the decision interval, it is said to signal that a potential change
• A quality engineer will then begin to search for the specific cause of change
• We assume that human sentiment can be considered as a stochastic process

Reviews various methods of statistical process control

Cumulative-Sum Control Chart ? good at detecting small changes in mean over time, built-in change point detection, calculate transform value for each time period, two charts (to detect increase and decrease)

Enron: Data

Enron was a company that traded energy like the stock market

After a major scandal, the email records of the company were made publicly available

We investigate the 50,000 emails sent from people within the company to others in the company

Identified 11 significant time periods ?

Salient character strings identified for 4 change pts: Enron, swap/counterparty/agreements/terminate/meters/ectcc. (That’s not a typo, it’s actually the salient character string.)

Can look at this and say, hey the organization has changed, these are the terms that have changed, then can search for cause by searching for emails that have a high ranking for these terms

Enron Results: There were 36 emails identified by the TFIDF salient terms within three weeks of the CUSUM signaled weeks. Of these 36, 14 suggested a possible cause of change:

The research provides a method of combining TFIDF with SPC to investigate change in sentiment over time. Future research in sentiment analysis includes:

• Applying additional classification algorithms to ENRON
• Applying this approach to other data sets
• Applying against real-time data to predict organizational change