March 31, 2011

Mid point..Lessons learnt

This post will keep on updating as and when I find new stuff and I realize something that I should have done.., etc



  1. Read as many as research papers possible..i have gone through 125+ so far..:(

  2. Don't try to group the research papers and put them in separate folders, because one paper may have to be put into several folders then. (If you try you will end up getting a headache ;) )

  3. Rename the research papers to the name of the research.(windows might not let you have long names, but linux does.., so go for linux :P)

  4. Use Linux (to name the papers really long and help with dolphin, specifically tabs is amazing..or you could install KDE on windows)

  5. As told above put all research papers in one folder and number them. I used to number as "01_RP (name)" where RP means research paper, BK for book and AR for articles

  6. Maintain an excel sheet with the paper number, name of paper, authors(useful when referencing) and tag the papers according to its content(example i used TOK for tokenization, ENC for encoding.., etc). it is important that you save these data because incase you loose the files you can get help of google easily. and  i assure that time spent on this will not go wasted.

  7. Another tip for skimming through the papers is to, first read the abstract and see how relevant it is to your project, you can even colour code the excel sheet depending on their importance. for example, for me papers on Sinhala were the most important so I colour coded it. But don't use more than 2-3, because then it looks highly unprofessional and complex.

  8. Also check the references for the papers you skim and search for them and read...etc..etc..it keeps going..that's how I ended up with 125 ;)

  9. So when u want to write about tokenization check the papers with TOK tag and read them.

  10. You don't have to read the whole paper to tag them. just skim through the headings, and highlight important stuff as you go through them. Okular would be a good option for reviews on pdf.

  11. Maintain only a single document. Don't try to break the doc to the modules or cores of your research, it will create inconsistencies. Make sure you keep the navigator window open when you are writing.

  12. Remember to keep back up always. Keep online back up as well. I have uploaded all my research papers to Google docs. All 125 of them ;)

  13. Finish doc as soon as you can. Leave at least 4-5 days to go through all the research papers again. Trust me, you will find loads more facts you have missed. It makes the doc complete because you will be thorough with the subject by then and will be able to grasp facts more quickly.

  14. Don't get worried if you feel like it takes too long (given that you have enough time for deadline), because that means you are getting it into your nerves.

  15. Always think how you actually do something, think "programming". Don't just put something on doc, put exactly what you will be doing, and in detail. If you feel like programming something is too much work, go for an API. It is very important that you think in programming level and will help you a lot later on. And I suppose anyone would have knowledge and experience to think deep. Think of the trouble you had earlier. Think of where you got stuck earlier. Think of where you could get into trouble. After all it's FYP, you should be able to do this ;)

  16. And finally your output should be professional. And here's where you should head back to windows, because Microsoft office is the best office package and to my knowledge no one has beaten it yet. This is where you must polish(format) your doc.

  17. When you do citations don't try to apply the Harvard guidelines at once. Now that you have numbered your research papers use numbered citations (e.g. [12]..). It is easy when documenting. If you have filled out the excel sheet I mentioned, then citing with Harvard is not difficult afterwards.


Mind you, that this is what worked for me. It might not work for most, but have a go if you think it will be helpful.. :D

No comments:

Post a Comment