# HG changeset patch
# User masa
# Date 1455800382 -32400
# Node ID 94ec38d9bdc6b805f101f5cd8b4fc0af1d33cff0
# Parent  f58bbc4a42f8146b99b1e93d89c64e0dd5da1fbe
add abstruct eng

diff -r f58bbc4a42f8 -r 94ec38d9bdc6 paper/abstract.tex
--- a/paper/abstract.tex	Thu Feb 18 21:39:33 2016 +0900
+++ b/paper/abstract.tex	Thu Feb 18 21:59:42 2016 +0900
@@ -8,31 +8,26 @@
 それぞれの例題によって結果の整合性を取る必要があるが、どのように整合性を取るかは問題によって考慮する必要がある。
 
 本研究で実装した正規表現は、正規表現から正規表現木を生成し、その正規表現木に状態を割り振りながらNFA を生成する。
-もし NFA が生成した場合は Subset Construction で DFA に変換する。
-そして、DFA の生成後、ファイルとマッチングさせる。
+NFA を  Subset Construction で DFA に変換する。
+そして、ファイルとマッチングさせる。
 
+マッチングはマッチの Range と次の状態の組の配列(tState)を生成し、tState を状態遷移させておこなう。
+並列処理はブロック単位で行うが、ブロックにまたがる Pattern はあとでやり直すというシンプルな方法を採用した。
 それぞれの例題と Mac に built-in されている wc、egrep と比較し測定を行なった。
 \end{abstract}
 
 \begin{abstract_eng}
 Cerium is a parallel programming framework, which developed by our laboratory.
-We implemented file read with mmap in parallel string search.
-It is easy to use mmap system call to read from file, but current implementation of mmap sometimes does work well.
-So we implement `Blocked Read', that is reading file separated by a certain size.
-
-Examples of parallel string search are included in Cerium.
-They are Word Count, Boyer-Moore String Search and Regular Explession.
-
-It needs to check for correct result in devided file every examples.
+We implemented parallel file processing using both mmap and multi threaded read.
+The multi-threaded read uses Blocked Read. The parallel proccesing is performed for each block as a map task.
+The results of map tasks are summed in a reduce task.
 
-Implemented Regular Explession sequential execute.
-At first, convert Regular Explession to binary tree.
-Second, allocated state to binary tree.
-Third, convert state with Subset Construction.
-Finally, implement parallel regular expression.
+Examples of parallel string proccesing in Cerium are including Word Count, Boyer-Moore Search and Regular Expression matching.
+Regular expression matching uses DFA generation (Subset Construction).
+The matching is performed by array of tState, which are pair of matching range of characters and next tState.
+The parallel proccesing of the matching is per-block based.
+In case of matching patterns which acrosses blocks, the matching processes are executed in the reduce task.
 
-We evaluate examples of Cerium, wc of Mac built-in, and egrep.
-
-
+We compare examples of Cerium with UNIX wc and egrep.
 
 \end{abstract_eng}