DHAKA: Executive Director(ED) of Bangladesh Computer Council (BCC) Shwapon Kumer Sarkar was the chief guest in a seminar titled ‘Use of Bengali language in Information Technology’ (IT). The key- speech was presented by former ED of BCC S. M. Ashraful Islam.
The seminar was moderated by BASIS president Mostofa Zabbar. At the very beginning he said that this seminar is historical because have been discussing about use of Bengali language in IT. Some people have become successful and advanced. At private initiatives today there is no problem in writing Bengali mechanically. It was a challenge for us. Earlier Bengali was written in type machines now it is written in computer. In last January a project of Taka 159 crore two lakhs has been approved. The project includes the work of develop 16 tools. Today we will discuss how this project can be implemented.
In his key-note speech SM Ashsraful Islam mentioned the problems of using Bengali language in IT and added that our problem is fount and another is key-board. Again to translate is a major problem. To understand Bengali hand writing is a problem. Auto-spell checker is a problem in Bengali. Bengali Unicode invention has removed the problem of fount. In key-board Bijoy has become very much popular simultaneously Avro key-board mixing with other key-boards has brought Interface. Still there is anarchy remains with key-board. There is still major problem with translation. We had to research a lot on integration of IT with Bengali language. There are two ways in natural language processing. One is classical and another is empirical and we are to proceed with these two. Previously corpus was made of paper then the size was problem but now the size is no problem.
In this project we will prepare many features of Bengali language now it will be shaped to standardize. Next we will go for OCR which has been made by ICT division and it will be developed more. We will make Speech to text and Text to speech. National key-board will be made usable. We will further make Bengali spell and grammar checker like English. Besides, technology will be developed for disable people. Moreover we have a plan to include Ahamia because it is a part of Bengali language. We will proceed step by step.
Enamul Kabir said that the content of Bengali in Internet is less than 0.1%. If we cannot increase the volume content in Internet then it will not be sustainable. We discuss with Unicode regarding full-stop and double full-stop. This is owned by others. We want our own code of Bengali language. So to implement these we are to work a lot. To do all these we will need technical expertise and language expertise. We also need a skilled manager.
Samira Juberi Himika said it was our dream that on the online there is sufficient Bengali content. We started our work with this concept in 2012. After searching a lot that we saw in Shahjalal University there was some progress was already done. Next we started doing work on double OCR. After some work for one and half year we handed over the specification to the government. We want to help the government in development. We have finished more work. We will release API very soon so that people can use it free. In implementing the project those who work on artificial intelligence globally we use their experience. We want that llet us share the work with competition.
We need software engineer. In this field the help of industry is needed. It took 70-80 years in making English OCR so it will be not practical to do it in 3 years. However if we can work together then it will take less time but it needs commitment.
Dr. Hassan Sarwar said since 2005-6 we have been doing with Bengali of my own. From A2i program we were given Taka 23 lakhs for manufacturing OCR. More work is going on for its development. One cannot do all work at a time. We need researchers. More man power is needed for software architect ring and engineering. The goods which will be sold in the market it needs procurement innovation.
Faruqe Ahmed of A2i said there is some difference of previous OCR and present OCR. We need to develop both the OCR. For that we will be required to do some research work. Along with motivation we need capable developer. We are sincerely working to involve these works with A2i project. It will be better to prepare TOR through formation of some open groups.
Nurunnabi Hasib said that to make any language permanent it should be on the on line. So we have been working for a long time with Wikipedia for the development of Bengali language. Through Wikipedia so far 47 thousand Bengali articles have been included. There are complains against us that we have wrong information about Bengali. We are trying to do better. We are to use Bengali in the android phone by installing Bengali fount. We have a team named Wiki source. We voluntarily can work with Bengali language. If the youth is involved then they can contribute a lot. Image to text can be included in the project.
Munir Hassan said that about 50 lakh Bengali words are needed by using Google Static Database. Goggle has 24 millions Bengali words of which 14 millions we supplied. If the place corpus can be made more standardize then it will be work.
If we can divide our work then the project implementation will be easier.
AFM Daniul Haque said that Initially Munir, Shahid, Bijoy, Avro, Mayabi, Android and Parag of India work should be united. Same course should be followed for Bangladeshi Bengali and Kolkata Bengali. We could not proceed accordingly because of skilled man power. There some hassle in Joint Bengali fount. So long government had no initiative. In January 2017 government has taken up a project of Taka 159200000/-. For the successful implementation of the project public opinion should be created.
Enamul Kabir said that recently we got approval of the project. However 3 year time is not enough to Bengali ICT platform. The content Bengali in the on line is below 0.1 percent. So if we fail to put our footprint correctly the digital transformation is difficult. There are nine Unicode for nine languages. We are to work a lot in Unicode. For this we need man power.
DR. Mansur Musa said that about 99% percent Bangalee do not what is Bengali language. If they don’t have grammar then many things of the language will be incomplete. Bengali language on the basis of region there are variations. The language of computer is limited. The machine will not translate beyond the limit. This needs some theoretical work. In this case we should first decide the domain. I am hopeful that this time government has com forward. We can depend on the young generation. Awareness should be created among teachers and students of the schools. I call upon the government, semi-government and private sectors should work jointly. We can also call upon the on line group to work in this sector.
EC of BCC Shawpon Kumer:
Those who are working behind this project, I would request them BCC alone is not responsible for this project rather it is duty of our all. So you will also be hurt if it fails, I believe. The matter of RFP is also under consideration. We still could not appoint any manpower for doing wok in this project. For integration of manpower I need the help of every body.
BDST: 2115 HRS, MAR 29, 2017
SJN/AP
