- 138
- 235 018
Computational Linguistics
Germany
Приєднався 6 тра 2019
The purpose of this channel is to communicate the basic terminologies and their implementation related to text mining, natural language processing and sentiment analysis in specific and data science and machine learning in general. We will also discuss and implement codes for social networks analysis and their structural representation through graphs and then processing those graphs for analysis.
Відео
Simple Graphs
Переглядів 208Рік тому
The video explains what are the properties of a simple graph. It explains the mathematical representation of graphs. It further explains the rules that ensure that a graph is a simple graph. Such rules are a) not having loops, b) no multiple edges between the same pair of vertices, c) the graph is undirected and d) the graph is unweighted. Simple graphs are more often used are they are computat...
Complex network analysis
Переглядів 1 тис.Рік тому
Complex network analysis refers to the study of large networks that possess properties which could not be found otherwise in simple graphs. They generally represent bigger systems, like the networks of all web pages on the WWW. It has emerged as a new field comprising basics of graph theory, community detection and machine learning. These networks can be studied in static or dynamic arrangement...
Social network analysis
Переглядів 1,1 тис.2 роки тому
What is social network analysis, network analysis and graph theory. Social network analysis helps understand public behavior or response towards commercial products, government policies or any trending issue. It has been used as the first source of information in emergency situations and natural disasters. Analyzing social networks data for in-time and better decision-making is becoming more of...
data cleaning (Course Promo)
Переглядів 622 роки тому
Data cleaning is where data scientists spend the majority of their time. The data is coming from multiple sources and may not sit well together. Furthermore, there may be human errors and errors due to software/hardware failure. If the data received from the real-world sources is not properly cleaned, scaled, organized and well presented, even the latest models may not yield good results. It is...
Handling Missing Values and Noise Values (Univariate Outliers)
Переглядів 1582 роки тому
The lecture explains how to deal with missing values and noise values or univariate outliers in a dataset. It is part of the data cleansing or cleaning series that discusses different techniques for detecting and handling various issues in data. These tasks are performed as preprocessing the data so that only good quality data can be provided for training machine learning models. The techniques...
feature types
Переглядів 3682 роки тому
There are two types of features in structured data. They are numerical features and categorical features. Numerical features are represented by either Integers (discrete numbers) or real (floating point) values. Categorical features are represented by string values. They may either be Ordinal or Nominal. Ordinal categorical features are string values with a specific order, while there is no suc...
Language basic analytics (Roman Burushaski)
Переглядів 572 роки тому
The video provide code walkthrough of basic analytics of burushaski. It identifies some statistics about the dataset like max sentence size, word size, average words per sentence and few others. It also shows how to generate the wordcloud for the dataset and performed topic modeling and its visualizations as well.
Endangered Languages (Burushaski)
Переглядів 702 роки тому
There are many languages that are endangered of going extinct. We can preserve these languages by building tools that would promote the digital use of these languages. Burushaski is an endangered language with 87000 speakers categorized as vulnerable by UNESCO. #languages #endangeredlanguages #Burushaski
Text Preprocessing, NLP
Переглядів 3392 роки тому
The video provides theoretical explanation and coding exercises for the various preprocessing steps used on textual data. Which preprocessing steps are required on your data depends on the nature of tasks you intend to perform. More straightforward tasks like classifying between two types of class labels may require more preprocessing to reduce computational cost. The information lost in prepro...
RM L4: Writing Literature review, proposed work and experimental results
Переглядів 492 роки тому
RM L4: Writing Literature review, proposed work and experimental results
RM L3: Reading / Writing Abstract, Introduction and Literature Review
Переглядів 622 роки тому
RM L3: Reading / Writing Abstract, Introduction and Literature Review
NLP L4: Text Vectorization and Text Categorization
Переглядів 1492 роки тому
NLP L4: Text Vectorization and Text Categorization
WP L3: Web Technologies and Web application architecture
Переглядів 432 роки тому
WP L3: Web Technologies and Web application architecture
WP L2: History of Internet and Worldwide web
Переглядів 152 роки тому
WP L2: History of Internet and Worldwide web
WP, L1: Overview and outline (Urdu / English)
Переглядів 272 роки тому
WP, L1: Overview and outline (Urdu / English)
Link Analysis: Betweenness centrality
Переглядів 15 тис.3 роки тому
Link Analysis: Betweenness centrality
Network Analysis: Link Analysis using closeness centrality
Переглядів 1,2 тис.3 роки тому
Network Analysis: Link Analysis using closeness centrality
MLDS MID Q5 Answer: Calculating Precision and Recall from Confusion Matrix
Переглядів 7 тис.3 роки тому
MLDS MID Q5 Answer: Calculating Precision and Recall from Confusion Matrix
Thanks SiR.
how you could say Burushaski is endangered. It is not.
Great 👍👍👍
👍👍👍
not understtood
in 0/1, how will you get 1?
Thank you
I love you
Muito bom, excelente !
I have a skepticism about the betweenness centrality in this lecture cause the CB(3) in undirected condition would be 6 but you mentioned it as 0.6. I hope you will mend it.
How n=6 and not 7?
Good video, however you should have a double ended arrow between 3 and 4. Thank you
tnx man u saved us
cool
notes??
Why we don"t involve the situation of j>k in the directed graph when we calculate the betweeness centrality? I can understand we ignore these situations in the undirected graph, because the result is double . In this directed graph, if we calculate E→C， the result will be different. I mean E does have an effect on C in the directed graph, which is not reflected in the paths we have already calculated, right?
thanks! it helps a lot!
How calculate f1score?
Thanks a lot 😊
Thanks for the tutorial
Best vedio I've seen so far thank you soo much
hello! can i ask what book are you using ?
Thanks :)
Hold up... isn't the closeness centrality of a node the inverse of the mean of the shortest paths between it and all others? So, wouldn't C(A) be 4/(1 + 2 + 2 + 3)? Meanwhile, the normalized value divides it by (n - 1), effectively making C'(A) = 1/(1 + 2 + 2 + 3), right?
For those wondering the type is actually the normalised closeness centrality measure ie the closeness centrality divided by the total number of nodes -1. Since the pre normalised centrality is the sum raised to the -1 we get the type on screen
Really helpful!! thanks
You’ve saved my life I swear❤️
detailed explanation with multiple example 🔥🔥 great
How to find betweeness of 1000 nodes in undirected graph?? Please do reply for this
thank you
Great explanation. Thank you
Thank you for your presentation. But,in case of many nodes,calculations can be difficult. Based on your video, I show how to calculate prestige in Rstudio. This is the link. ua-cam.com/video/LFVLvYMPmaI/v-deo.html
Thanks , for this educational contribution.
Thank you sir
BEST TEACHER!!!
why not indegree was considered?and when to consider indegree and outdegree for degree centraility?
Great work sir, really useful stuff and the way you explain it! Awesome!
Glad you think so!
Sir How can I get complete code that you wrote
Dear Irfan, it's advised that you code along watching the video and better have your own modifications / experimentation with the code to understand better.
how there is 6 in numarator for node 3
Appreciate the videos!
well explained :)
well explained
in directed graph why we consider only A to C distance and not C to A?
Yes, from C to A should also be considered as they will be different paths. The same applies to DA, EA, DC, EC, ED also. However, in the example, all of these values (DA, EA, DC, EC, ED) will come out to be zero and will not affect the actual result. DA will not be included as there will be no path from D to A.
@@771aryanand there is no path from C to A so we have ignored 👍🏻
@@lakshsinghania bro j<k so we wont consider those paths
Thanks for the video. It was educative but I noticed you did not calculate the degree of centrality for node 2 GREAT VIDEO!!
Node 2 will also have degree centrality of 0.4 (in case the graph is undirected).
Made sense
actually really useful, thanks!
Glad to hear!
Some segments in the video are stamped not adjacent to each other
You teach really good👍🏻 everything explained so well👌🏻
Glad you think so!
Can you please upload the ppt in the description link
Simple and easy explanation.