本文選題：問(wèn)答系統(tǒng) + 詞向量　； 參考：《浙江大學(xué)》2017年碩士論文

【摘要】：問(wèn)答系統(tǒng)是目前自然語(yǔ)言處理領(lǐng)域中的研究熱點(diǎn),它既能讓用戶(hù)通過(guò)自然語(yǔ)言直接發(fā)問(wèn),又能直接向用戶(hù)返回精確、簡(jiǎn)潔的答案,而不是一系列相關(guān)網(wǎng)頁(yè)。近年來(lái),深度學(xué)習(xí)技術(shù)為問(wèn)答系統(tǒng)領(lǐng)域帶來(lái)諸多突破,基于深度學(xué)習(xí)技術(shù)的問(wèn)答算法研究成為了自然語(yǔ)言處理最熱門(mén)的研究方向,誕生了大量?jī)?yōu)秀的文章與開(kāi)發(fā)框架,如Google在2016年推出的SyntaxNet,大大降低了高性能問(wèn)答系統(tǒng)的開(kāi)發(fā)成本。本文應(yīng)用了深度學(xué)習(xí)技術(shù)進(jìn)行問(wèn)答系統(tǒng)的構(gòu)造,開(kāi)展的工作如下:1.利用詞向量與卷積神經(jīng)網(wǎng)絡(luò)搭建了一套高準(zhǔn)確率的面向具體任務(wù)問(wèn)答系統(tǒng),改進(jìn)了已有的卷積神經(jīng)網(wǎng)絡(luò)問(wèn)句分類(lèi)算法,探索了模型初始化參數(shù)與模型性能的關(guān)系。2.基于雙向長(zhǎng)短時(shí)記憶模型與注意力機(jī)制搭建了一套端到端開(kāi)放領(lǐng)域問(wèn)答系統(tǒng),改進(jìn)了前人基于單向長(zhǎng)短時(shí)記憶模型的端到端問(wèn)答算法在問(wèn)句語(yǔ)義表征上的缺點(diǎn)。3.在Facebook bAbI、Ubuntu Dialogue Corpus等常用數(shù)據(jù)集上進(jìn)行了實(shí)驗(yàn)對(duì)比,通過(guò)實(shí)驗(yàn)結(jié)果對(duì)比突出了本文設(shè)計(jì)的問(wèn)答算法的有效性與合理性,并對(duì)實(shí)驗(yàn)結(jié)果做了較為詳細(xì)的分析。4.利用TensorFlow、Docker構(gòu)建了一套維護(hù)成本低、部署方便的問(wèn)答系統(tǒng)微服務(wù),解決了 TensorFlow框架線上服務(wù)部署困難的問(wèn)題。本文的主要貢獻(xiàn)如下:1.創(chuàng)造性地發(fā)現(xiàn)了基于詞向量和卷積神經(jīng)網(wǎng)絡(luò)的問(wèn)句語(yǔ)義相似度算法性能與詞向量維數(shù)之間的關(guān)系,并通過(guò)實(shí)驗(yàn)加以驗(yàn)證。2.嘗試了通過(guò)復(fù)制插值的方式擴(kuò)展基于詞向量和卷積神經(jīng)網(wǎng)絡(luò)的問(wèn)句語(yǔ)義相似度算法中詞向量輸入部分的維數(shù),解決了問(wèn)句類(lèi)別數(shù)上升時(shí)模型性能下降的問(wèn)題。3.使用雙向長(zhǎng)短時(shí)記憶模型與注意力機(jī)制改進(jìn)了現(xiàn)有基于循環(huán)神經(jīng)網(wǎng)絡(luò)的端到端問(wèn)答算法模型,提高了平均問(wèn)答長(zhǎng)度等性能指標(biāo)。4.基于TensorFlow與Docker實(shí)現(xiàn)了一整套問(wèn)答系統(tǒng)微服務(wù),創(chuàng)新性地使用Spring Boot包裝算法腳本,解決了TensorFlow Serving的兼容性問(wèn)題,實(shí)現(xiàn)了彈性部署與擴(kuò)容,維護(hù)成本低。
[Abstract]:Question and answer system is a hot research topic in the field of natural language processing. It can not only let users directly ask questions through natural language, but also return accurate and concise answers to users directly, rather than a series of related web pages. In recent years, deep learning technology has brought many breakthroughs to the field of question and answer system. The research of question and answer algorithm based on deep learning technology has become the most popular research direction of natural language processing, and a large number of excellent articles and development frameworks have been born. SyntaxNet, for example, launched by Google in 2016, has greatly reduced the cost of developing a high-performance question-and-answer system. In this paper, the deep learning technology is used to construct the Q & A system, and the work is as follows: 1. Using word vector and convolutional neural network, a set of quizzes oriented question answering system with high accuracy is set up. The existing convolutional neural network question classification algorithm is improved, and the relationship between model initialization parameters and model performance is explored. An end-to-end open domain question-and-answer system based on bidirectional long and short term memory model and attention mechanism is constructed, which improves the shortcomings of the previous end-to-end question answering algorithm based on one-way long and short term memory model on the semantic representation of question sentences. The experimental results are compared on Facebook bAbIbuntu Dialogue Corpus and other common data sets. The validity and rationality of the question and answer algorithm designed in this paper are highlighted by the comparison of experimental results, and the experimental results are analyzed in detail. By using Tensor flow Docker, a question-and-answer system micro-service with low maintenance cost and convenient deployment is constructed, which solves the problem of difficult service deployment on the TensorFlow framework. The main contributions of this paper are as follows: 1. The relationship between the performance of semantic similarity algorithm based on word vector and convolution neural network and the dimension of word vector is found out creatively. This paper attempts to extend the dimension of word vector input in the semantic similarity algorithm of question sentence based on word vector and convolutional neural network by replicating and interpolating, and solves the problem of deterioration of model performance when the number of question categories increases. The existing end-to-end question-and-answer algorithm model based on cyclic neural network is improved by using bidirectional long and short time memory model and attention mechanism, and the performance index of average question length is improved. Based on TensorFlow and Docker, a set of question answering system micro-service is implemented, and Spring Boot packaging algorithm script is innovatively used, which solves the compatibility problem of TensorFlow Serving, realizes flexible deployment and expansion, and has low maintenance cost.
【學(xué)位授予單位】：浙江大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP391.1

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王樹(shù)西;趙星秋;潘碩;;問(wèn)答系統(tǒng)在教學(xué)中的應(yīng)用[J];中國(guó)教育信息化;2007年07期

2 毛先領(lǐng);李曉明;;問(wèn)答系統(tǒng)研究綜述[J];計(jì)算機(jī)科學(xué)與探索;2012年03期

3 莫麗萍,王樹(shù)西,姜吉發(fā),雷雨霞;問(wèn)答系統(tǒng)和淺層結(jié)構(gòu)模式推理[J];廣西師范大學(xué)學(xué)報(bào)(自然科學(xué)版);2004年01期

4 盧志堅(jiān),張冬茉;中文問(wèn)答系統(tǒng)中的問(wèn)句理解[J];計(jì)算機(jī)工程;2004年18期

5 王樹(shù)西;問(wèn)答系統(tǒng):核心技術(shù)、發(fā)展趨勢(shì)[J];計(jì)算機(jī)工程與應(yīng)用;2005年18期

6 林曉慶;;問(wèn)答系統(tǒng)中基于列表類(lèi)問(wèn)題的研究[J];電腦知識(shí)與技術(shù)(學(xué)術(shù)交流);2007年07期

7 張積賓;徐志明;王恒;潘啟樹(shù);;面向大規(guī)模網(wǎng)絡(luò)數(shù)據(jù)的社會(huì)化問(wèn)答系統(tǒng)[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);2008年12期

8 賈君枝;毛海飛;;漢語(yǔ)框架網(wǎng)絡(luò)問(wèn)答系統(tǒng)問(wèn)句處理研究[J];圖書(shū)情報(bào)工作;2008年10期

9 胡小華;劉軒;劉丹;陸偉;;基于冗余的仿真問(wèn)答系統(tǒng)的輕量級(jí)局部文本分析[J];圖書(shū)情報(bào)知識(shí);2009年01期

10 張中峰;李秋丹;;社區(qū)問(wèn)答系統(tǒng)研究綜述[J];計(jì)算機(jī)科學(xué);2010年11期

相關(guān)會(huì)議論文 前10條

1 何靖;陳，

本文編號(hào)：1960603