Try the book "Introduction to Natural Language Processing Application Development in 15 Steps" --Chapter 1 Memo "Preliminary Knowledge Before Beginning Exercises"

This is a memo for myself as I read Introduction to Natural Language Processing Applications in 15 Steps.

Preparation

--Personal MacPC: MacOS Mojave version 10.14.6 --docker version: Version 19.03.2 for both Client and Server

Chapter overview

As described in the book, we will cover the construction of the learning environment and the knowledge (programming language and library) that are the prerequisites for the exercises.

Note

1 Introduction / Natural language processing and machine learning

No notes

2 Writing / development environment for this book

The writing environment is distributed as requirements.txt, and Dockerfile is also distributed.

The requirements.txt describes the type and version of the library as shown below, and you can install the library in bulk using pip.

`requirements.txt`


numpy==1.15.0
scipy=1.1.0
...
mecab-python3==0.996.2
neologdn==0.3.2
...

#Installation example with pip
$ pip install --no-cache-dir -r requirements.txt

When using Dockerfile, build and start Docker container from Dockerfile. (I run it here)

###Container build###
#Run in the location where the Dockerfile is located
$ ls | grep Dockerfile
Dockerfile

#Build with repository name: 15step, tag name: latest
$ docker build -t 15step:latest .

###Container startup###
#When the container is started by default, python is started
$ docker run -it 15step:latest
Python 3.6.5 (default, Jun 27 2018, 08:22:23)
>>>

#Stop is quit()
>>>quit()
$

#How to start python after starting bash
$ docker run -it 15step:latest /bin/bash
/usr/src/app# 
/usr/src/app# python
Python 3.6.5 (default, Jun 27 2018, 08:22:23)
>>>

3 Python basics for machine learning

--Allows you to use lists, tuples, and dictionaries properly --Tuples are immutable ――How to write ternary operators in English compared to C language - price = 1000 if age >= 20 else 800 --If age> = 20, price = 1000 is executed, otherwise price = 800 is executed. --for statement --For each list, for val in <list>: --If you want a counter variable, set enumerate to for i, val in enumerate (<list>): --If you want only the counter variable, set range to for i in range (5): --Functions, classes, modules --Load module ʻimport module as md --Call a function withmd.method --from module import method to load module functions --Call a function with method --Regular expressions can be executed faster if the regular expression pattern is compiled in advance.

4 Numerical library NumPy

--The array is numpy.array (<list>) --The dimension is <array> .shape --Slice (obtained by specifying multiple elements) is <array> [m: n] --Get elements m ~ n-1 --Optional if m = 0 or n = --Broadcasting --Automatically aligns dimensions when calculating vectors + scalars and arrays + vectors --The four arithmetic operations can be calculated with the <array 1> operator <array 2> --The inner product is numpy.dot (<array 1>, <array 2>) --If you want to get the maximum value --Getting the maximum value of an element is numpy.max (<array>) --Getting the index of the maximum value of an element is numpy.argmax (<array>) --Total, average, random number, initialization (initialize with 0, initialize with 1) --Sequence join --Vertical direction is numpy.vstack (<array 1>, <array 2>) --Horizontal direction is numpy.hstack (<array 1>, <array 2>)

5 Other major libraries used in this document