Coincidences

It begins with Pramoedya Ananta Toer's "Saya Terbakar Amarah Sendiri" when I came across the writer's name: Andre Vltchek. After that, it's an array of coincidences. The movie: Terlena, The Breaking of a Nation. This article: Natural Disasters or Mass Murder?. All of him.

things that can become research agenda

• orientation-based evaluation metrics : output alone is inadequate but complete with alignment. Good translation has a good reordering, good reordering has a good orientation
  
• phrase-boundary model : Suppose there are three phrases A B C D and A is a function word, A may inversely reorder B, B C or B C D under no penalty except language model in current model. How about letting the boundary's orientation (B C for B, C D for B C, and D for B C D) to judge it
• Non-frequency based preference model : Again, suppose there are A B C D where B and C are function words and their orientation is already effective to their neighbor (B to A and C to D). At higher constituent, where should we pick as head (B or C)? We need a head preference model. Currently we just reward that with higher frequency. Is it nicer if we have P_pref(B|C) and P_pref(C|B) learned from training data.
  

kecil dan besar

Dari sebuah milis :

kecilnya dicerca igit-igit damprat, tuanya suka ngeyel dan mengutuk

kecilnya tertelikung panasnya kekerasan, tuanya demen tengkar perang

kecilnya terolok-olok, tuanya maluan dan memalukan

kecilnya didorong bimbing, tuanya tabah ksatria

kecilnya dipupuk toleransi, tuanya sabar dan bijak

kecilnya dihargai, tuanya tak benci memuji

kecilnya berlimpah pemahaman dan penerimaan, tuanya mengerti kasih

kecilnya tak kekurangan pengakuan, tuanya akan tahu dan sadar diri

kecilnya dididik kejujuran, tuanya dapat dipercaya

kecilnya merasa aman, tuanya dapat bergaul diri dan orang lain

kecilnya diajari keramahan dan adil, tuanya dapat mengerti kedamaian

(kiriman Dorothy, 1972)

kecilnya bermasalah, tuanya selalu bikin masalah
kecilnya kurang perhatian, tuanya ngeyelan
kecilnya tahu gotong royong, tuanya bukan sektarian
kecilnya mendendam, tuanya cari-cari kesalahan lingkungan
kecilnya bodo, tuanya tak bisa membedakan ruang privat dan publik
......

"John loves Mary" or "Mary loves John"

I believe that a man is meant for a girl and that's true for the other around, just like John loves Mary implies (hopefully) Mary loves John. However, that can't be true for "John kicks Mary" which doesn't imply anything near Mary kicking John.

In many SMT systems, language model plays a very big role in determining correct translation. Researchers at Google with virtually unlimited data can use the greatest-language-model-ever-built to filter out unrealistic translation. But imagine this scenario when you are translating a simple sentence and the top two best output of translation model are : "John loves Mary" and "Mary loves John". Or "John kicks Mary" and "Mary kicks John". All are grammatically correct, but there's no way the language model can tell which one is correct.

In this case, language model is useless and a little peek at source sentence would be a great help.

PS: May I change John with my name and Mary with her, so that she can love me too?... hahaha

thinking in a scrambled order

Let me tell myself about my research. If I translate a Chinese sentence to English, I care not only about having the correct word translation but also the correct ordering. I am currently investigating the latter problem of reordering. PS: we call it reordering and not ordering because the objective is to reorder the sentence from the Chinese to English order. Oh, by the way, my research is statistical machine translation.

If I enjoy my first paragraph, then there must be a problem with the way I think. Ok, maybe "must" is a little bit too strong of a word. But if I analyze it closer, I will notice that I should have said something earlier to make it nicer.

My research is statistical machine translation, in particular I am investigating reordering problem. Let me be more specific about this problem. If I want to translate a Chinese sentence to English, I should care not only about getting the word translation correct but more importantly arranging them in an order that will be readable as a fluent English sentence. We call it reordering problem because usually the task is to reorder the Chinese-ordered translation to English order.

I admit I make a lot of ordering mistake in making my point across. I start with a lead with the hope of making people to understand what I'm saying but end up confuse them even at the very start. It is often the case that I am not sure about the point I want to make. When I write this paragraph, I edit it five times permissible in writing but not in oral presentation. When I challenge my believe about my point, I suspect that I automatically erase the prepared argumentation flow but only to find that my point is actually correct.

I think I have to increase the reordering accuracy in my way of thinking from now on, not only in translation quality.

PS: This is me talking to myself, so the pronoun is first person reflective...

only

Ada indah, ada cerlang
ada merdu, ada harum
Semua ada, tapi hanya belaka
The only real thing is her

Engkau panglima bermuka dua !! Omie Lubis

  Engkau tak pantas berhiaskan bintang dipundak mu
 bila semua dosamu harus kupikul di pundak ku
 Engkau tak pantas bersujud mencium sejadah
 bila kemunafikan mu harus kusangga
 Engkau tak pantas memanggilku sebagai anakmu
 bila kau tak pernah mencintaiku
 
 Seharusnya di lenganmu aku bergelayut
 seharusnya dalam senyum mu aku meraih mimpi
 seharusnya dimatamu aku memuja
 ketika aku merasa lelah dan putus asa
 seharusnya dipundak mu aku bertahan
 ketika hempasan badai tiba meronta
 
 Senjatamu bukan untuk melindungiku
 tapi untuk membunuhi saudara saudaraku
 kekuatan mu bukan untuk melindungiku
 tapi untuk melumpuhkanku
 
 kau sambut tangan musuh
 kau tepis rangkulanku
 kau suapi mulut musuh
 kau tampar mulutku
 kau sanjung musuh
 kau hina dinakan aku
 
 Panglima macam apakah kau ini????
 
 
 
 
 puisi ini kutulis untuk saudara2 ku di tanah air
 jiwa ku akan selalu bersama mu !
 
 Omie lubis