?

Log in

Little Sasi's Little Thoughts

Recent Entries

9/20/08 02:46 pm - Sutra - an effort at building a localisation support tool...

Localisation is still a relatively unknown thing in India, despite its linguistic diversity, and large number of people who cant read/write English. One hopes to see more activity on this front, in the days to come! We tried our hands in localisation for a while in the early days of OSSRC (www.ossrc.org.in) and even compiled a guide for people on how to localise software. Thence came the idea of a tool to support localisation of software as it passes through version changes. Sutra was the result. It took about a year and half (low (wo)man power investment) to get into a usable form. It is now accessible through OSSRC portal and www.cdacmumbai.in

Localisation is an interesting domain for trying out a lot of natural language processing ideas. NLP is not doing great in open domains like Web, due to the inherent richness and complexity of language in unrestricted environment. Therefore, NLP is looking for useful, but restricted domains for effective performance. Localisation is one such area. The language is usually formal and focussed. Mostly all text associated with a software fall into one domain. One does not often see much fancy writing. At the same time, it is not too trivial either - there is adequate sentence complexity, and also a lack of large relevant corpus. Thus, it offers a non-trivial domain for automatic lexicon updation, translation support, etc. Some of these are in the future plan for Sutra.

Sutra effort is from the Open source group here at CDAC Mumbai, and is led by Aparna. You can read a paper on that in the September issue of E-gov magazine. The link is valid only if you are reading this in September, otherwise look for the September issue in the archives of the site. In a companion article in the same issue, you can see more of my thoughts on NLP and localisation.

9/16/08 07:59 pm - Workshop on expert systems at Navi Mumbai....

I now have a lot of news to share on many fronts... Starting now with an upcoming event at our campus...

A 2-day workshop on expert systems is coming up at Kharghar campus during October 17-18, 2008. Expert systems used to be a major focus area in the AI group at NCST/CDAC in the late 80s and early 90s, with the expert system shell Vidwan being one of our achievements. It was available to people at a throw away price and became fairly popular in the expert system community in India then. It was the only ES shell developed in India and available to public. Starting from its birth on a Unix platform in Prolog, it moved to C on DOS, then to C on Unix (using Curses package!), then to Windows, and now it is available as a Web application. The project was dormant for a few years, and is now active in porting, enhancing, etc. It is a good time now to reintroduce ideas like expert systems into software applications to boost up their intelligence and analysis power. As applications move to the next level of complexity, these would become necessary ingredients. We possibly wont see many people using specialised languages like Prolog for such work, but the ideas and frameworks are likely to play significant roles.

The workshop will introduce the field of expert systems (focus on rulebased representation), and use Vidwan for the demonstrations and practicals. It will be of interest for those interested in developing applications such as troubleshooting (equipments, people(!), processes, etc), selection (of tools, products, etc), and so on in various areas such as medicine, finance, technology, etc.

For more details, refer to our website.


8/22/08 07:59 pm - Symposium on Education and Technology in schools...

I am in Bangalore last 3 days attending this symposium organised by Quest Alliance in partnership with US AID, Intel education, Azim Premji foundation, etc. I was cynical about the programme, and on the spur of the moment wrote to the organisers (in response to their call for participation) criticising the attempted import of foreign wisdom to transform our school education. After a round of a warm (not hot) debate, I decided to attend, and they offered to pay for my travel!

There were a fair variety of people in the gathering, though mostly foreigners only were the main speakers. Prof Jalaluddin from India gave the keynote - a nice overview of many aspects from history of introducing computers in Indian schools (from Mrs Indira Gandhi's time, thanks to a visit by Queen Elizabeth), the major developments since then, the challenges including content, etc. The other speakers were Dr Nancy Law from Hongkong, Punya Mishra (from Michigan State Univ - who was an invited speaker in our Vidyakash conference at NCST about 6 years back), and Dr Kozma (dont remember from where) and Mrs Geetha Narayanan from India. The talks were generally good and informative - but did not connect with each other and left the symposium with no specific directions or conclusions at the end.

It was nice to know that there are lots of institutions looking at the use of technology in school education including corporates like IBM, Intel and Oracle, a number of NGOs, some of the schools themselves, etc. There is good amount of information to share - but sadly, these were never given the dais to showcase. I think we would have learned more by critically looking at their successes and failures, than looking at what has been done in US.

I was part of a large panel on the first day - a crude attempt to gather the state of ICT in education in India - and got 5 minutes to comment on technology trends in this space. I used the space to pitch for open source, and richer shareable localised content. It turned out that the 2.5 min pitch on open source was well taken by a lot of people, going by the number of people who met me later on for more detailed discussion. It is a nice sign to see the interest from the practioners in open source and open content - it augurs well for the growth.

Punya Misra did a live blogging of the event, and you can find it here. Quest Alliance is planning to put up all the resources, videos, etc on their website shortly!
Watch this video - an interesting story.

8/14/08 08:21 pm - Naivasha Lake, Kenya

This one is stitched together from three separate shots - utility provided by the camera software.

8/14/08 08:08 pm - Pictures from Kenya...

Here are a few pictures from Kenya - see story given in earlier blog. They are behind a cut not to tax your browser, unless you want to see the pics.

See picturesCollapse )

8/6/08 08:27 pm - Hi from Kenya....

Went to Nairobi, Kenya for a project meeting on 5th evening, along with Ankit and Mayank. The purpose was to finalise the requirements for building a system like Mulyaankan for Kenya. Expenses for the trip was taken care of by Kenya. The nice surprise was that they arranged business class tickets for all of us - perhaps, because the Indian customs valuation commissioner was accompanying us. It was the first business class travel on international flight for me. We got lounge access at the airport (which provided some relief to missing the lunch), good near-100% reclinable seats in the flight.  Food was nothing great - but ok. Carrier was Kenya airways. We were in Kenya from 6th to 10th and reached back Mumbai on Monday morning. Kenyan Revenue Authority (KRA) was the host for the visit.

Nairobi - KRA headquarters, and the base for our meeting - is a laid back city like Delhi. Americans have structured the city landscape like Newyork with a lot of skyscrappers buildings. No sign of any local culture in the city heart infrastructure. The transport was amazingly slow. Apparently, the police blames the vehicle for any case of hitting a pedestrian, and hence any vehicle would draw to a stop the moment you step into the road - it does not matter, if there is a zebra crossing or not - making the overall traffic movement very very slow. The city is relatively expensive, despite being a third world country -- comparing currency value with purchase power against rupee --. Food is mostly non-veg. The local folks cant think of a meal without meat - so, the veggies in the group had a hell of a time.

On Saturday, we managed to visit "Hell's gate", near the Naivasha lake, about 100 km from Nairobi. Enroute were huge -- running into kilometers in length and breadth -- flower farms for rose, etc. The entry is through Elsa gate where you take the ticket, and you can drive inside (walk at places) or hire a bike for riding. As you move to the hills/gorge, you can see zebras, giraffes (pictures coming soon). You stop after a few kilometers, and take to foot. It is a steady climb down through jungle stream, narrow pathways through rock, following a stream. About 30 minutes down, you can find really hot streams -- capable of burning your hand, The place is pretty clean, and not much crowded. It is a place worth a visit. Youtube has some videos (not by us) on this tour - example1

The hot water down under is exploited to generate electricity - first time, I am encountering geothermal electricity generation. They produce about 100 MW of electricity from this area itself, with the water under running to over 300 degree C. It is also an area prone to vulcanos; no recent large eruptions, but small ones are often seen (so they say!).

7/31/08 05:45 pm - Birthday post...

This time for the birthday, I happen to be in NCST Bangalore. I had forgotten the birthday, till Sawani came in the morning and wished.

As has been the practice for the last couple of years, here is the greetings count.

In person: 2 (add another 10 or so, thanks to the cake cutting function)
On phone: 2
SMS: 3
E-mail: about 10 (mostly Kharghar folks)
Orkut: 35

Orkut is clearly the winner this time. I guess partly because not too many people in Bangalore knows me enough to worry about my birthday.

The day started with finding the following mail from Prof Sivakumar, IIT Bombay, which was a nice surprise. "sasi- just saw your review of a Semantic web book as the Editor's Pick at http://www.reviews.com/  congrats. cheers,"  I have reviewed about 10 books and many papers for ACM Computing reviews, and it is nice to have a review tagged as "Editor's pick" and highlighted on the homepage.

Among the e-mails wishing me was a mail from Kingfisher offering the usual wishes, and guess what they had a gift too. They put 75 miles into my frequent flier pool [If this is repeated for the next 100 years, I may get a free ticket to Bangalore from Mumbai, only on this basis!]. Anyway, nice gesture.

And then in the evening, the LCG folks in Bangalore arranged the cake cutting too. Our ED happened to be around (for the same meeting which brought me to Bangalore) and he also joined in. I did manage to escape a cake-facial this time too.

- Sasi

7/28/08 08:36 pm - Computing curricula and ACM

While searching for something, landed up on an initiative ACM has been pursuing for the last few years -- started in 2002, it says -- to propose a curriculum for computer courses at undergrad level. [Even the Americans can take ages to come up with a syllabus, eh? ]

The effort is quite extensive in two ways. They have identified 5 related disciplines and defined them fairly well on what they focus on. These are computer engineering (more hardware/architecture focus), computer science (more theory focus), information technology (more application focus), information systems (higher level application design), software engineering. Their CC overview report shows the picture well. For each, there is a committee prescribing what a graduate of that stream should know, what are core components and what are optional, fairly detailed concept level breakup of the courses, and a lot of guidance on how to adapt them to individual requirements.

One may or may not agree with their specific course recommendations, but there were things which tallied with our own thoughts with respect to courses, and hence found nice. One is the continued emphasis on core programming skills along with the new theme of integrative programming. This is important, more so, since a recent issue of CACM had a view column saying core programming is no longer required for IT professionals, "hardly any jobs for C/C++ programmers", he says. Of course, another person countered the view point well in the same issue (CACM July 2008). Second is the significant emphasis on Human computer interface - it is a core topic in the curricula and covers a lot of topics including multimedia interfaces. Third is the emphasis on professional ethics, communication skills, team work, integrity,  and the like. A good many pages of the 130 page report deals with the importance of this and how to cover these as pervasive topics through the entire curriculum. Those parts are worth a read for every IT professional. And the entire report if you are into teaching for IT students. URL for report.

- Sasi

7/21/08 01:18 pm - Pentagon!

Pentagon


I have five corners; they are my outlets to the world.

Each feel they represent me the best,

Understand the best in me and nurture for the rest.

Little do they know me, for what I stood.

“Times have changed and so we must change”

I hear them repeat, everytime they meet.

When you lose your core values,

do they know, that we cease to exist?

Perhaps nothing matters any more;

in a world that is after fame and money.

The route matters not, just the end;

My hard-earned respect dies at the end.


The corners are so funny, in many a colour;

“so colourful” you could perhaps say,

“so ugly” the combination looks to me.

Some have no corner, bending with every breeze;

Some are so full of holes, hardly a corner!

I feel tired, hopes dried up as rivers in summer;

Years I have waited, my eyes are swollen.

Will there be a corner that I can love with my heart,

that truly understands me and my broken heart?


Update: The "I" here is not the little sasi, and actually not any person -- living or dead. It is a personification.

7/10/08 04:45 pm - Difficult Misconceptions...

Here are a couple of points where, if a student answers wrong, you should not completely blame them. Even the books sometimes compound the confusion.

1. What parameter passing mechanism does C use? call by value, call by reference.

Most people answer this as "both". How does one get 'call by reference'? By passing the address of the argument, of course! One does not realise that the address is now the actual parameter and it is passed by value. So, the answer should be that only call by value is supported.

2. Does Java support multiple inheritance?

The answer one often gets is: "no, but we can use interfaces to implement multiple inheritance". The rationale is that a class can implement any number of interfaces, just as in multiple inheritance. But an interface has nothing worth inheriting. Implementing an inheritance only makes objects of this class eligible to stand in for objects of the interface type. So, use of interface has nothing to do with multiple inheritance - except in a very narrow sense of allowing the class to match the classes represented by the interfaces.

- Sasi

6/22/08 08:10 pm - A photo-blog

Archana, dressed to the occassion, among like minded people near Chowgule College, Goa.





A view of nature

6/18/08 02:26 pm - Dijsktra and today's IT scenario...

Here are some excerpts from an article written by EWD many years ago. Much of it applies so much today too. The article is EWD920.

The computer industry, however, operates under the equally tacit assumption that —thank the Lord!— computing science has nothing to contribute to its business, which —as everyone knows— is primarily a matter of marketing, of timely inventing the proper slogan in name of which the next castle-in-the-air can be presented as solid real-estate.

The university departments are in trouble too: the computing boom has decreased their staff and increased their teaching load, and as far as computing science is concerned, most faculty members have no time for it anymore and most students are no longer interested in it.

[Academic community] has to refine and to teach to the best of its abilities how computing should be done; would it ever yield to the pressure to propagate the malpractice of today, it had better fold up.

If the current malpractice continues too long, the computer industry will collapse and it is hard to save an industry that is no longer there.

6/7/08 08:06 pm - Memories from an interview...

Been in some interviews for CDAC-Mumbai recently. I won't give more details, so as to protect the guilty! Here are some amusing pieces, with a little adaptation of text.

Case 1:
[after some unsuccessful attempts to get some sensible responses...]
member1: How many bits are in a byte?
Cand: [lights up face - first question he can answer] 8
member 1: good. How many bytes in a kilobyte?
Cand: 1000
member 1: 1000 bits or bytes?
Cand: [after thinking a bit] bytes.
member 2: are you sure it is exactly 1000?
Cand: [more thinking] yes.
member 2: [offering a clue] That is wrong. You are almost right, though.
Cand: [more thinking] 999, Sir.
...Panel shocked for a minute, and has a good laugh....
member 3 tries more, asking him to write down 2, 2^2, 2^3, etc. But he does not get the clue.

Case 2:
member1: why do we need normalisation? [candidate has indicated DBMS as his favourite subject]
Cand: [After a lot of thinking] it helps to remove redundancy.
member 1: why do you want to remove redundancy?
Cand: redundancy is bad, creates problems.
member 1: I think redundancy is good. It helps to recover data when there are errors. It is good to keep multiple copies.
Cand: [looks confused] But redundancy creates problems.
member 1: what problems?
Cand: it increases congestion.
member 1: how? you are accessing things only from one source, though there may be multiple copies. So how does it affect congesion?
Cand: [more confused] That is what the book says.
[We decide to continue the interview with the author of the textbook instead...]

6/7/08 07:50 pm - A tale of two airports...

Been to Hyderabad and Bangalore recently through their new 'international' style airports. Both share a lot of similarity. Both are bulit mostly out of glass. From the aircraft, you can almost see the people checking in, particularly in the night. I dont like such heavy glass structures - it is too much glass. And it is very confusing to find your way, in the midst of the see-throughs and reflections. Huge parking space (and an exhorbitant parking charge of Rs 50 per 15 minutes at Bangalore), good transport arrangements with buses, taxis and a vast pick-up range, etc are other commonalities. The really bad part is the distance from the city. Worst is Bangalore for this - it is about 70+ km from E-city, the primary IT hub. If you have an evening flight, you better pack up in the morning after breakfast. Driving time ranges from 90 min to 150min, ie more than the flying time from most places to Bangalore!

Both places badly need a fast dedicated metro-like train service from airport to city. Will someone build an "Indian" airport large enough, convenient, spacious and reflecting Indian environment? Kochi airport was something like that - looked like a typical Kerala palace. But it is already running out of space, and expansions are in progress, and God knows which "modern" style is coming there!

5/25/08 11:16 am - MaTra is flying!

MaTra, our English to Hindi machine translation system, is now online in a new avatar with better performance. Thanks to the young MaTra team who has taken on the challenge of revamping and re-engineering the old and largely fragile system. A number of engineering decisions and improvements have exposed a lot of the power hidden in the system. Some real enhancements are also in progress. At present, it does simple sentences with a fair degree of richness. Basic structure for compound/complex sentences are present - but still has some problems. Similarly, questions, etc are also not handled properly.

Here are samples:

Machine translation is a very difficult and complex problem.
मशीन अनुवाद एक बहुत कठिन और जटिल समस्या है।

Only idiots work on machine translation.
केवल गधे मशीन अनुवाद पर काम करता है। [why did it miss the plural verb?]

MaTra team is very dedicated.
मात्रा टीम बहुत समर्पित है।

Is not this wonderful?
यह अद्भुत हैं। [Of course, Matra is intelligent, it knows what it is doing! So it has no doubt.]

Read and more try it out at: MaTra

5/14/08 09:46 am - Top ten algorithms of the 20th Century

When looking for something, bumped into this interesting thread. IEEE Computing Science puts up a list of top-10 algorithms formulated in the 20th century. Given that computers really came on the scene around the middle of the century and led to tremendous interest in formulating programmable algorithms, most of the top-10 are from the middle of the century. Most of these are unlikely to be familiar to most computer science students, I think. I should I look up and understand them. Not that the top-10 is defined based on its impact on the field of science and engineering, and not on the beauty or elegance of the algorithm. Hence, these are not necessarily popular in a usual algorithms class. You may also not necessarily agree with the selection. Anyway, it is nice to see quick sort sitting pretty among all those high-sounding algorithms.

Here is the list in one-liner form. And the link below for those who want more details or proper reference.

 1946: Monte Carlo Method
1947: Simplex method for linear programming
1950: Krylov subspace iteration methods
1951: Decompositional approach to matrix computations
1957: Fortran optimising compiler
1959-61: QR algorithm for computing eigen values
1962: Quick sort
1965: Fast fourier transform
1977: integer relation detection algorithm
1987: fast multipole algorithm

Links:
IEEE Explore - summary
Another editorial introduction

4/29/08 08:06 pm - CACM celebrates 50 years...

CACM - communications of ACM - is celebrating its 50th birthday with an anniversary special issue of January 2008. It is worth a look, for those who have not stopped reading completely! It has some nice reflections on the evolutions of the magazine and associated ACM transactions, some reflections on the past and future of computing, etc. As is usually the case, CACM articles are not hardcore research with difficult terms and even more difficult equations - they are light to read and gives you a feel for the area/topic.

It is amazing to look at what computing has gone through in 50 years. I still remember our first programming work was on what was called a BBC micro with 64K RAM. From there to 640K (old DOS systems) to MBs to today 2 GB RAM look at the scale in a span of less than 25 years! Those were days internet was something hardcore academic, e-mail was largely unknown in India, there was no Web to talk of, graphical user interface was too expensive (only the Mac computer had anything like that), object oriented programming was yet to be born - forget component technologies, and the like. Today, it is hard to imagine a computer world without all these. Today, we have no problem compiling and running a factorial program a hundred times in a day to get it right - back then, it would have costed you a minimum of 3 months! From the punchcard days, when we moved to an 'interactive' (read dump) terminal from where we could edit a program (line editor only) and submit directly - it was the arrival of heaven on earth. Today, there is no worse hell than that! A 2GB machine with a grand IDE with one click compilation and error detection is too painful. On the positive side, we knew how to program (even without a machine at hand) in those days and we knew how the computer would behave when given a program. Most computer programmers dont know this today - their programs (yes, the source code) is written by the compiler or the IDE by trial and error. To know what a 4 line program does, he/she needs the computer (with an IDE and GUI!).

3/31/08 08:04 pm - A trip to Orissa

Got a chance to visit Orissa - Bhubaneswar last week for a workshop organised by Alka's team on language literate computing (whatever that means!). I gave a talk on open source software and an overview of natural language processing. With a number of people talking in Oriya (which I dont understand!) and much of the remaining talks wandering aimlessly here and there, the technical component of the programme was uninteresting. However, I could manage a visit to Konark and Puri in the process. Here is a view of the Konark temple. Due to security restrictions, no photo from Puri. We also visited a few Siva temples in the city. All show exceptionally detailed carvings on rock.

Due to being night time, etc the few photos I took did not come out well. So, this is pretty much all there is to share.

When we reached Puri temple it was about 9pm, and had to wait till 11.00pm for the temple to re-open (apparently Lord Jagannath was having a shave!). The temple is full of money-hungry priests, who trick you into parting with money very easily. I have seen similar hunger in Varanasi. It is sad to see these people exploiting such places like this.

3/23/08 02:13 pm - Some Photos...

Everybody seems to be in a photo-mode. Let me add my share to that also. Here is a small set. I am putting my collection up on photobucket.com

Nature - a nice view of the river near Rasayani, Mumbai


A couple of sunsets: one from Nasik and one from Nanded.


3/21/08 03:43 pm - OOXML and Indian IT industry

India managed to keep its stand of voting against OOXML being an international standard with 13 votes to 4. The entire vote against came from the so-called 'indian IT industry'. I don't think they would care any bit about the standards - nor have they so far. If big American companies are unhappy with them, they will cease to exist - their existence is so much like a soap-bubble. What else can they do, but yield to Microsoft pressures?

The number of problems identified by countries the world over, the dirty proceedings of the BRM, and the hasty manner in which the standards were pushed for acceptance in ISO and to top it all, the fact that MS accepted almost every problem pointed out as valid: all show clearly the sorry state of the proposed standard. And it is a pity that the stalwarts of Indian industry chose to turn a blind eye to these.

3/20/08 04:14 pm - Biased reporting and Businessworld...

Businessworld web title says "number one portal....." and today when searching for something bumped into this report: This report is completely biased. Particularly look at the last few sentences.

BIS has been evaluating both the standards for more than a year now. It had appointed a committee headed by D B Phatak of Kanwal Rekhi School of Information Technology for an assessment. Based on the committee's recommendation the BIS will say yes, no, or abstain on the vote for one of these standards to be accepted by the International Standards Organization (ISO).

At the August 23 meeting of the BIS committee, the Mr Phatak threw out 120 of the 200 objections to the acceptance of OOXML as a standard in less than 45 minutes. Still 17 of the 21 members of the committee rejected approving OOXML as a standard, confirming India's stand that there should be just one standard globally which would be ODF.

BIS already had a committee named LITD-15 headed by Ms Neeta Varma from NIC. In this particular meeting, Dr Phatak did play a key role in handling the discussion and controlling the temper. But no special committee was formed to take a decision. The decision was already taken in LITD-15 to vote "no, with comments". The subgroup spent a few hours in sorting out the comments available to choose the best 100 or so comments for the international forum. Not a single comment was thrown out. Given the limited time available, thanks to being the end of the day, the process was rushed. Even Dr Phatak said at the end that Microsoft must address ALL the comments, not just the one we are taking to ISO - of course, Microsoft conveniently forgot all that soon after.

India's stand was never that ODF will be the chosen standard. India was voting OOXML down as it was ill conceived, buggy and too hurried. That was all. Comparison with ODF was rarely an issue.

But this report in forums like Businessworld stinks. I can never trust any report they carry either.

What is even more interesting that the site offers a poll for users to choose ODF or OOXML, and ODF wins more than 90% of the votes there.

3/13/08 08:00 pm - Competitions....

I happened to be the chief-guest (more appropriately 'cheap-guest') at the concluding function of their annual festival for an engineering college nearby. It is always a bit embarassing to play this role, given my usual informal dress style, where one expects a tie-suit clad or a very elderly person for such roles. Anyway, I accepted on compulsion, and spoke a few words too -- apart from distributing prizes. The message I focussed was on the nature of competitions.

Most colleges, in their annual function, have the standard set of competitions: sport and cultural. Fashion seems to become a regular - perhaps since it provides an opportunity to break rules which may otherwise be imposed. Robotics is another thing I am seeing becoming popular. This is good and can help build something nice. They should have something like robotic club and work regularly, rather than putting up a contraption just 2-3 weeks before the event. That way, it can result in something nice.

I was suggesting focussing on competitions on solving socially relevant problems. Accessibility - making IT accessible to people with disabilities of various sorts - is a theme of concern for my group here these days. So I brought that up - how about building devices for the blind and other disabled to interact with the computer -- instead of the traditional screen and keyboard. In a recent trip to Nanded, I saw the theme of looking at IT in agriculture. This could be another interesting competition - innovations from technology to help the farmers in their various activities. Devices -- software and hardware -- to solve socially significant problems would be a good target.

I dont think my little talk had much of an impact - the audience was eager to get me out of the way, so that they can get their prizes and get lost in the dance programme which was to follow.

2/26/08 08:40 pm - Conferences, conferences...

It is raining conferences in India. Every other college is organising at least a national conference every year, most likely on something relating to information technology. And almost all of them claim submission of at least a hundred papers. If you run a good quality conference, it is hard to get even a dozen good papers. That shows the quality of papers coming in - it could vary from introduction to quantum computing to Wifi standards, and more often that cut and paste work other sources without even attributing credit. It is becoming a common genetic disorder now a days!

On the other hand, these play a good role too. It gives the average teacher an opportunity to talk to an audience different from their students. At least those who are serious would get some feedback to learn from, and get inspiration to do something original and substantial. It teaches them issues in organising a conference from proceedings to scheduling.

During Dec-Feb, I have been part of a number of such conferences - as a member of their advisory committee, an invited speaker, session chair, or a keynote speaker. The talks are normally popular - "sir, you gave a full talk on neural networks without a single mathematical formula" - and are often meant to generate some excitement about the subject and research in the audience, rather than sharing some research achievements. So I use common place examples - often picked on the fly - and focus on insight and ideas, rather than mathematical form or formal structure. This goes well with most of the audience, in general.

During the period I gave talks on accessibility, artificial intelligence, machine learning, e-learning, artificial intelligence and the web, parallel computing, etc. I enjoy these little outings - visiting relatively remote places and interacting with serious (but deprived of opportunities) people. I generally plan to spend a little time doing some sight seeing, but often end up spending the entire time at the conference venue itself.

2/26/08 07:43 pm - OOXML - a truly international standard....

Microsoft developed something like a standard to compete with ODF - yes, officially it is not Microsoft and it is not to compete with ODF, but that is the official version. It was one of the worst horrors in this direction. A 6000 page document forms the standard. All countries in the world were given 3 months to study this document and give comments. 3 months is 90 days (assuming no holidays incl sat/sun). So if one person is to understand the entire standard, he should read about 70 pages a day - impossible for any human being for 3 months at a stretch. The ISO standards body had no problem in imposing this punishment on the world, and the world yielded to a large extent. Most countries (e.g India) has had large bodies which discusses such issues, containing representatives from academia and industry. Each of the members are expected to study the document and come back with comments/observations.  Assuming about 100 countries studying this document to identify problems and offer corrections, look at the man days Microsoft gained free of charge to improve their document. Microsoft would have lost a fortune if it were to bear the salary and other expenses for this effort!

Fortunately, most of the world managed to find far too many problems in the document - an example of open source development methodology!! I hope Microsoft softens its hatred of FOSS after this! And the standards process was slowed down due to the large number of negative votes. More than 2000 significant comments were offered on the document.

Microsoft came back with another thousand+ page document containing responses to the comments. Again the responsibility of the world to study this and revise their stand. And again, they are all yielding for whatever reason.

Looking back, no other standard would have received so much of international participation in its making. Possibly no other standard was so buggy and baggy. Size does indeed matter - of Microsoft, if not that of the standard document. And so, the final outcome is not guaranteed to be a merit-based decision.

2/12/08 01:39 pm - Another legend passes away....

Dr RK Joshi, a pioneer in Indian language fonts and related aspects, passed away recently while on a visit to US to attend the Unicode consortium meeting. I remember him from the time I joined NCST, though given the large gap between our areas of interest, I had little direct interaction. He has been deeply passionate about his subject and his work. The excitement and contentment in the way he would explain some of his posters to visitors is worth watching. He made the otherwise lifeless fonts into works of art, and media of communication. He would organise and arrange letters from multiple languages to create amazing paintings - often with deep serious messages. His knowledge of the field is also quite amazing, and puts him in sharp contrast to those who pretend passion without enough depth to backup.

I wont forget the enthu he put in to make the first open house which we organised, a success. I realised how ordinary my contribution has been seeing what he has done going out of the way to visualise and plan things in which he is good at: Individual division logos modelled along the lines of CDAC logo, design template for all posters, leaflets, etc, a poster exhibition depicting NCST's history from inception till date, The second open house also he spared no effort to make it have that uniqueness stamp. The Vidyakash logo, Vivek journal cover, all our exhibition posters, etc all had RK's stamp of innovation.

He has been a champion of the phonetic encoding, (as opposed to the Unicode encoding) - a cause he pursued till the end. He has also been working on Sanskrit (Vedic form) font structure enhancing the current font template to support a five layer font structure.

May his soul rest in peace...
Powered by LiveJournal.com