小C的第一宇宙
基于TheBrain的知识分析 初探
wangc
May 11, 2018
阅读本文需要 14 分钟(按字数)

尝试用thebrain进行个人知识管理已经有一年半的时间了，这短时间来学习的知识点形成了一个超过1.8万个节点以及超过2.3关系连线的庞大的思维导图。在thebrain软件中存储了关于这些节点与连线详细的信息，于是尝试导出后用python进行一下简单的数据分析，作为进一年多来学习情况的一个总结汇报。

数据读取

import glob
import codecs
from html import unescape
import calendar
import matplotlib.pyplot as plt
import numpy as np

# 文件名读取
filenames = glob.glob("thoughtData/*.xml")
for i in range(len(filenames)):
    filenames[i] = int(filenames[i][12:-4])
filenames = sorted(filenames)

# 文件数据读取
thoughts = {}
for i in filenames:
    thought = {}
    thoughts[i] = thought
    name = ""
    child = []
    jump = []
    parent = []
    filename = 'thoughtData/' + str(i) + '.xml'
    for line in open(filename):
        if ('name' in line) and (len(name) == 0):
            start = line.find('name') + 6
            end = line[start:].find(' ') + start -1
            name = unescape(line[start:end])
            continue
        if 'modificationDate' in line:
            start = line.find('date') + 23
            date = unescape(line[start:-2])
        if 'Link' in line:
            str1 = 'idA="'+str(i)+'"'
            str2 = 'idB="'+str(i)+'"'
            
            start = line.find('idB') + 5
            end = line[start:].find(' ') + start -1
            start2 = line.find('idA') + 5
            end2 = line[start2:].find(' ') + start2 -1
            
            if ('CHILD' in line) and (str1 in line):
                child.append(int(line[start:end]))
            if ('JUMP' in line) and (str1 in line):
                jump.append(int(line[start:end]))
            if ('PARENT' in line) and (str2 in line):
                parent.append(int(line[start2:end2]))
    thought['name'] = name
    thought['modDate'] = date
    thought['child'] = child
    thought['jump'] = jump
    thought['parent'] = parent
```python


## 数据处理

帮助函数

def compareDates(date1, date2): date1 = date1.split(‘-‘) date2 = date2.split(‘-‘) for i in range(3): if int(date1[i]) > int(date2[i]): return ‘>’ elif int(date1[i]) < int(date2[i]): return ‘<’ return ‘=’

时间序列处理—转换成二十四小时制

modDates = [] modTimes = [] for i in filenames: modDate = thoughts[i][‘modDate’] if ‘上午’ in modDate: modDate = modDate.replace(‘上午’,’’) modDates.append(modDate) elif ‘下午’ in modDate: date = modDate start = date.find(‘下午’) time = date[start+2:] end = time.find(‘:’) hour = str(int(time[:end]) + 12) time = hour + time[end:] modDate = date[:start] + time modDates.append(modDate) start = modDate.find(‘ ‘) end = modDate.find(‘:’) modTime = modDate[start+1:end] modTimes.append(modTime)

修改时间序列

for i in range(len(modDates)): end = modDates[i].find(‘ ‘) modDates[i] = modDates[i][:end] len(modDates)

耗时！

创建时间序列

createDates = []

for i in range(len(modDates)):

date = modDates[i]

for j in range(i, len(modDates)):

if compareDates(date, modDates[j]) == ‘>’:

date = modDates[j]

createDates.append(date)

每日想法统计

dateCount = [] dates = dateGenerator(2017,7,10,Y,M,D)

for date1 in dates: count = 0 for date2 in modDates: if date1 == date2: count += 1 dateCount.append(count)

sum(dateCount)

每日想法统计

dateCount1 = [] dates = dateGenerator(2017,7,10,Y,M,D)

for date1 in dates: count = 0 for date2 in createDates: if date1 == date2: count += 1 dateCount1.append(count) sum(dateCount1)

每小时想法统计

timeCount = [] times = range(1,25) flags = len(modTimes)*[0]

for time in times: i = 0 count = modTimes.count(str(time)) timeCount.append(count)

timeCount[22]=timeCount[22]-800

sum(timeCount)


## 数据分析

-- coding: utf-8 --

每日节点统计

modification = list(np.array(dateCount)-np.array(dateCount1)) for i in range(len(modification)): if modification[i] < 0: modification[i] = 0

创建+修改

dateCount2 = list(np.array(dateCount)+np.array(modification))

去掉导入1119个节点数的一天

for i in range (len(dateCount)): if dateCount[i]==max(dateCount): dateCount[i] = 100

plt.plot(modification,label=”modification”) plt.plot(dateCount,label=”creation”) plt.title(‘Day Count’) plt.legend() plt.show()

print(“节点总数：”,sum(dateCount)) print(“复习率：60%”) print(“修改率：”,round(sum(modification)/sum(dateCount),3)100,”%”) print(“未学习天数：”,round(dateCount.count(0)/len(dateCount),3)100,”%”) print(“平均每天学习：”,sum(dateCount2)//len(dateCount)) print(“最多的一天学习：”,max(dateCount))


![image](http://wangweiguang.xyz/images/pkma1.jpg)

节点增长

sums = [] for i in range(len(dateCount)): s = 0 for count in dateCount[:i+1]: s += count sums.append(s) plt.plot(sums) plt.title(‘Increase’) plt.show()


![image](http://wangweiguang.xyz/images/pkma2.jpg)

#一天中24小时的节点计数 plt.bar(range(1,25), timeCount) plt.title(‘Time Count’) plt.show()

![image](http://wangweiguang.xyz/images/pkma3.jpg)

整个知识网络的可视化

%matplotlib inline import matplotlib.pyplot as plt import networkx as nx import pydot from networkx.drawing.nx_pydot import graphviz_layout

画出子树

root = 4 # 数学 cildTree = findChild(4, list(G.edges))

ncolors = [‘b’]*G.number_of_nodes() for i in list(cildTree): ncolors[list(G.nodes()).index(i[0])] = ‘r’ ncolors[list(G.nodes()).index(i[1])] = ‘r’

ecolors = list(G.edges) for i in range(len(list(ecolors))): if ecolors[i] in cildTree: ecolors[i] = ‘r’ else: ecolors[i] = ‘b’

pos = graphviz_layout(G, prog=’dot’, args=’’) plt.figure(figsize=(20, 20)) nx.draw(G, pos, node_size=5, alpha=0.5, edge_color=ecolors, node_color=ncolors, with_labels=False) #plt.axis(‘equal’) plt.show() ```python image