[译] [709] 本章小结

[cssmagic]

Summary

本章小结

• We need to divide a big programming problem into smaller sub-problems before we can effectively implement it.
• Top-down design is a systematic technique for breaking a problem down into small subtask functions.
• In top-down design, we seek small functions that solve well-defined tasks and that can be used by one or more other functions.
• Author identification is the process of guessing the author of a mystery book.
• We can use features about words (e.g., average word length) and sentences (e.g., average number of words per sentence) to characterize how each known author writes.
• Machine learning is an important area of computer science that investigates how machines can learn from data and make predictions.
• In supervised learning, we have some training data in the form of objects (e.g., books) and their categories (who wrote each book). We can learn from that data to make predictions about new objects.
• A signature consists of a list of features, one signature per object.
• When we’re ready to implement our functions that arose from top-down design, we implement them from the bottom up; that is, we implement the leaf functions first, then functions that depend on those leaf functions, and so on until we implement the topmost function.
• Refactoring code means to improve the design of the code (e.g., by reducing code repetition).

• 在有效实施一个庞大编程任务之前，我们需要将其划分为更细小的子问题。
• 自顶向下设计是一个系统化技巧，用于把复杂问题拆解为多个小的子任务函数。
• 在自顶向下设计策略中，我们寻求那些能够解决具体任务并能被其他函数复用的小型函数。
• 作者识别是推测一部未知书籍作者的过程。
• 通过分析词语（例如，平均词长）和句子（例如，平均每句话的词数）的特性，我们能够描绘出每位已知作者的写作风格。
• 机器学习是计算机科学中的一个重要分支，它研究机器如何通过数据学习并进行预测。
• 在监督学习场景下，我们有一些训练数据，这些数据以对象（如书籍）及其分类（例如，每本书的作者）的形式存在。我们可以利用这些数据来对新的对象进行预测。
• 一个签名包括了一组特征，每个对象配有一个独特的签名。
• 当我们开始根据自顶向下的设计准备实现我们的功能时，我们采用自底向上的实现方式；也就是说，我们首先实施最基础的叶子功能，然后是依赖这些叶子功能的更高层次的功能，依次类推，直至实现最高层的功能。
• 重构代码旨在提升代码设计的质量（例如，通过降低代码冗余）。