siri控制智慧生活（剑桥小哥哥李泽辉）

Ever wondered why Siri seems to be getting funnier or online translations now appear startlingly accurate?

"One of the most important reasons is that Artificial Intelligence (AI) is now better at understanding human language," says Zehui Li, a Year-4 student of Computer Science with AI. He will study Natural Language Processing (NLP) - a branch of AI - at Cambridge this coming autumn.

(Plese scroll down for more details.)

想调戏手机里的智能助手，结果反被Siri怼到无语凝噎

语言完全不通，也能靠着翻译软件与国际友人social自如……

“人工智能越来越聪明，其中很重要的一个原因就是它们对人类语言的理解能力越来越强。”宁波诺丁汉大学计算机科学与人工智能专业大四的学生李泽辉说。

机器和人类语言会碰撞出什么样的火花？为了探索这个问题，他曾尝试开发了许多与自然语言处理有关的应用，并凭借一款“智能面试工具”获得全国编程大赛“黑客马拉松”第三名。现在，他即将前往剑桥大学，在人工智能的道路上继续探索。

李泽辉最早接触到自然语言处理是在大一暑假，“宁诺给我们提供了很多去海外夏校交流学习的机会，我选择了丹麦的奥胡斯大学。”他在夏校中遇到一位数学与人文社科兼修的指导老师，瞬间就被这种多学科交融的魅力所打动了。

“老师带我做的一个科研项目，是将丹麦图书馆里文学领域的藏书进行电子化处理，然后通过编程让计算机大批量提取和分析文本，我们再依据处理的结果，对从古至今文学作品主题的变化进行研究。”

这门计算机科学与语言学的交叉学科让李泽辉着迷，他对自然语言处理越来越感兴趣，并发现这项技术在现代社会中的应用已经非常广泛了，“我们平时用到的翻译器、能跟人对话的智能机器人、微信上的语音文字转换功能、Siri这样的语音助手，都是自然语言处理（Natural Language Processing，NLP）的具体应用。”

李泽辉希望在NLP上进行更深入的学习和研究，所以目标瞄准了这个领域处于世界领先地位的剑桥。“剑桥计算机学近年约30%的论文发布都集中在NLP。”他说。

为了顺利进入梦想中的名校，李泽辉做了充分的准备。“我所学的专业和之前做过的项目都是跟NLP比较相关的，但计算机对数学的要求非常高，所以我整个大三基本都花在学习数学上了。”他在大二、大三都拿到了均分80 的好成绩（在英国评分体系中，40分及格，70分以上为最高的一等学位）。

“面试之前，剑桥给了一个参考书目，需要阅读被指定书中的第十章。我把整本书都仔细读完了，指定的第十章更是翻来覆去看了不知道多少遍。面试时老师问了十多个专业问题，我都能回答到点子上，可能跟准备比较充分也有关吧。”李泽辉回忆道。

大二时，李泽辉和同学组队参加面向全国高校的编程界“才艺秀”——黑客马拉松，开发了一款能帮助HR提高工作效率的“智能面试工具”，这个工具能把应聘者面试中的语言转化为文本，并根据文本中的高频词、关键词对应聘者能力、性格特点进行分析。最终，这款“智能面试工具”让他们小组赢得了全国第三名。

之后，他又与同学参加了由微软举办的“编程之美”大赛，从1100多支大学生队伍中过关斩将冲入全国20强。

大三时，李泽辉前往英国，在学习之余找了一份实习，在巴克莱银行担任数据分析师，帮助银行建立起一个客户贷款信用评价系统，由于表现优秀，还获得了毕业后继续留在银行的Return Offer。但一心想要继续NLP学习的他，最终选择了告别同事和领导们。

“喜欢计算机，因为觉得这是一门实用性学科，竟然可以通过代码这种既抽象又摸不着的东西，来解决人们工作、生活和娱乐中具体存在的问题。”李泽辉很喜欢这种“技术改变生活”的感觉。

李泽辉(左)和朋友们

“如果有什么能称之为经验的东西跟学妹学弟分享的话，我觉得就是要积极主动地探索自己喜欢的东西，不要被外界的压力和‘随大流’的想法推着走。”李泽辉说。

(Please scroll down for more details.)

Merging language and AI for a smarter future

Ever wondered why Siri seems to be getting funnier or online translations now appear startlingly accurate?

"One of the most important reasons is that Artificial Intelligence (AI) is now better at understanding human language," says Zehui Li, a Year-4 student of Computer Science with AI. He will study Natural Language Processing (NLP) - a branch of AI - at Cambridge this coming autumn.

“I like the practicability of computer sciences – solving all-round, everyday problems through intangible codes,” says Zehui. Driven by curiosity, he has been actively participating in programming competitions (such as ‘Hackathon’) while at UNNC and has won national prizes with his classmates. During his exchange period in the UK, Zehui also did an internship as a data analyst at Barclays Bank, where he helped in setting up a customer credit evaluation system. This earned him an offer from the company to return as a full-time employee.

During a summer break in college, Zehui had his first experience of interdisciplinary research. “Among numerous ‘study abroad’ opportunities, I chose Aarhus University in Denmark and fortunately met a multidisciplinary instructor of both the mathematics and humanities and social sciences fields. I was stunned by the charm of such a combination of subjects.”

The research project contains both digitalisation and programming of library collections and subsequent analysation of literature texts, he said.

Gradually, Zehui found himself in love with NLP, which already has real-world applications. “Translators, intelligent robots, speech-to-text conversion tools and virtual assistants like Siri that make our life easier, all involve NLP,” says Zehui. “And it requires knowledge of both computer sciences and linguistics, which is irresistible to me as well.”

Knowing where his passion lies, Zehui decided to pursue a deeper understanding of the NLP field. “I am choosing Cambridge because it is particularly strong in this field in recent years. Almost 30% of its published papers on Computer Sciences are about NLP,” says Zehui. “I spent almost all my third year on mathematics, as a base for my future study of programming, and before the interview at Cambridge I read the whole reference book several times, especially the designated chapter.”

“If I have any experience to share, I think it is to always strive for your passion regardless of the so called ‘correct’ mainstream or other external factors,” says Zehui.

本期幕后

文字来源 | 高之寒

编辑整理 | Nico

English | Elena Yang

图片 | 李泽辉

排版 | WEAVER PR

©Copyright

转载或进一步了解

宁波诺丁汉大学品牌与传媒办公室