• 検索結果がありません。

パーサベンチマーク用言語Celloの設計と評価

N/A
N/A
Protected

Academic year: 2021

シェア "パーサベンチマーク用言語Celloの設計と評価"

Copied!
1
0
0

読み込み中.... (全文を見る)

全文

(1)情報処理学会論文誌. プログラミング. Vol.10 No.3 14 (June 2017). 発表概要. パーサベンチマーク用言語 Cello の設計と評価 山口 真弥1,a). 倉光 君郎1,b). 2016年10月26日発表. ベンチマークはソフトウェアやハードウェアを評価するための指標として広く用いられており,様々な 分野で標準化が行われている.しかし,パーサジェネレータにおいては標準化されたベンチマークは存在 しない.その理由としては,ベンチマークのテストデータとして利用するソースコードは多く存在するも のの,それらをパーサで解析するために必要な文法を開発することが困難であることと,パーサジェネ レータの評価として重要な項目である線形時間・空間性を評価するテストデータの収集が難しいことがあ げられる.本発表では,パーサジェネレータを評価するためのベンチマークを提案する.我々は実在する 大量のソースコードを解析し,その統計情報をもとに (1) 形式文法で定義しやすいベンチマーク用プログ ラミング言語の設計を行い,(2) テストデータをスケーラブルに自動生成する.これによって他の言語で解 析を行った場合の結果が予測しやすいベンチマークを作成する.また,開発したベンチマークと実在する ソースコードを用いて様々なパーサジェネレータを評価した結果を比較し,分析を行う.. Cello: A Simple Programming Language for Parser Generator Benchmarking Shin’ya Yamaguchi1,a). Kimio Kuramitsu1,b). Presented: October 26, 2016. Benchmarks are widely used for evaluating software and hardware systems. Since the importance of fair evaluation, several industrial benchmarks are standardized in order to ensure its fairness. However, for parser generator, there is no standard benchmark despite the existence of various parsing algorithms. Although there are much amount of source code as test data for benchmarking, it is involve significant cost to develop a grammar which is essential to generate a parser for the test data. Another reason is difficulty of correcting test data that cover range for testing scalability, which is an important point of view to assess capability of parser generators. In this presentation, we present ParserMark, a scalable benchmark for evaluating arbitrary parser generators. ParserMark is based on the statistics collected from a huge amount of source code in real world. By using the statistics, we (1) design a simple programming language for benchmarking, which expresses test data, and (2) generate test data that have the same tendency as real workload in arbitrary size. We evaluate several parser generator by using ParserMark and real workloads, and compare the results.. 1. a) b). 横浜国立大学 Yokohama National University, Yokohama, Kanagawa 240– 8501, Japan [email protected] [email protected]. c 2017 Information Processing Society of Japan . 14.

(2)

参照

関連したドキュメント

When we consider using WEKO as a data repository, it is not easy for the users to search the data which they wish because metadata are not well standardized in many academic fields..

Thirdly, we found that, using Christoffersen’s (1998) test and López’s (1998) loss function to compare models that have correct conditional coverage, we found that the

The relative commutant of B 1 in pAp is a direct sum of simple inductive limit algebras, each of which contains a simple real AF algebra (with the same K 0 group) and

For the survival data, we consider a model in the presence of cure; that is we took the mean of the Poisson process at time t as in (3.2) to be for i = 1, ..., 100, where Z i is

Using the results of Sections 1 and 2 one can also define in the same way as in 3.4 the set of isomorphism classes of “double” degeneration data associated with the minimal

As we have said in section 1 (Introduction), using the mentioned tree T , Barioli and Fallat gave the first example for which the equivalence between the problem of ordered

We presented simple and data-guided lexisearch algorithms that use path representation method for representing a tour for the benchmark asymmetric traveling salesman problem to

In this section we state our main theorems concerning the existence of a unique local solution to (SDP) and the continuous dependence on the initial data... τ is the initial time of