Skip to main content

Gallery

In order to now have too many blog posts with my drawings, I have decided to create a sort of gallery to put all my pictures on the same place.

So far it does not look like gallery, but it does have pictures, with dates and content written below it.

You can see it here: https://sarajaksa.eu/2019/gallery

My Master Thesis: Shrugging of New Ideas

I finally started working on my cognitive science master thesis. Yes, I am a sort of procrastinator. In a way, that I will always find something else to do, and finish other projects, but I will let some of them just continue for years.

So, what I am going to research (in case it will not blow in my face)? What I want to know is, what kind of network structure and individual differences impede and encourage the spreading of information. Sort of in a way, is the person willing to accept the new information. And because I had no idea, what kind of data to use, I have decided to try to do this on the fanfiction (Archive of our own) sites and with fandom.

I have tried to get some of the data, and it is possible. I just hope they don't block me, if I take more data from them.

I think this would be an interesting study, simply because I do want to know, what makes people believe what the do. And fanfiction communities are a good example of how the new knowledge, that a lot of people agree with, is created. I really wonder if I am just convincing myself?

Well, I still need to ask the other potential mentor, with whom did I already talked about. But that was months ago and I don't know if he is still interested. So this will be fun... (not).

And maybe this will help my explain, why I can shrug ideas easier than other people. I mean, beside that I learned how to live with my Ne, which is that I just ignore most of the ideas it throws at me. Maybe there is something in my personality dispositions or what kind of people do I hang out with, that makes it easier.

If anybody is asking (probably not), yes I had a conversation with somebody about my ability to shrug off ideas.

So yes, finally going forward with this as well. I really need to graduate soon and leave the structured academia.

World Value Survey, Slovenia and Justifiable Actions

When I was reading the article The boundaries of culture: do questions about societal norms reveal cultural differences?, I came across a very interesting passage for me (because I am Slovenian). This is the passage in question:

We noted that Slovenia exhibited some very unusual response patterns in 2005-2008. For example, unlike the respondents in any other country in the WVS, absolutely all Slovenes avoided the first four positions on the illegal-dishonest items. Therefore, we excluded Slovenia from our analysis.

This was the analysis of the following questions from the World Value Survey:

  • V198. Claiming government benefits to which you are not entitled
  • V199. Avoiding a fare on public transport
  • V200. Cheating on taxes if you have a chance
  • V201. Someone accepting a bribe in the course of their duties

That would me, that the people in my country avoided 4 positions in the 10-level measurements. That is almost half of measurement. It would means, that around 1000 of people that they asked, all of they has the same general direction of thinking.

World Value Survey does publish the summary information in their website. So I went and checked it. And what I found was not, that they avoided half of the answers. This are the answers:

Question Never justifiable 2 3 4 5 6 7 8 9 Always justifiable N
V198 402 125 136 68 44 104 41 21 7 22 1037
V199 412 91 122 74 50 103 34 42 21 25 1037
V200 565 135 97 49 30 49 25 19 10 18 1037
V201 699 109 68 29 17 33 15 10 5 10 1037

The numbers don't sum up, because of answered don't know or refused to answer.

Well, here it seems, that while it is skewed, the people in Slovenia used the whole range of answers. And I was so happy, that I would be able to find some proof in how corrupted with are in Slovenia (the idiotcracy of the stolen sandwich scandal not withstanding). Well, so much about that. It seems that most of them find it unacceptable. Around half of them find it unacceptable all the time. Which is completely different from the narrative, that is present in the Slovenian culture.

It also means, that not every detail can be trusted, even in scientific articles, without checking the data itself. Here it was easy, since it is publicly available, but what about when it is not? Plus, not everybody understand statistics enough to figure out something out of the data.

LSA on my Blog Posts

On my exchange in Bratislava, I had to take the Grounded Cognition class. The class was mostly the connection between linguistics, action-perception loops, computations and neurology.

In there, I had learned something about Latent Semantics Analysis (LSA). I can see, that unlike when I took the class, they had started with this. We had it in around the second third of the class. But the idea was interesting, and I figured out I could use it to find topics, that I write for in my blog.

Well, now it is a bot more than a year since then, and I decided to try and see, if I could do it. It was so simple, I could have just spend a hour or two doing it back then.

This is the code: LSA Clustering on my Blog Posts

Quora Kaggle Competition

Months ago, a friend of mine invited me to participate with him in a Kaggle competition. It was the Quora Insincere Questions Classification competition. I have to say, I am not even sure, we posted any model on the Kaggle on the end.

But I had used this as a sort of motivation. I needed to pass the linguistics class, so this seems like a good way to consolidate the knowledge and make linguistics as bit more interesting. Well, I passed the exam, though I don't know how much of this can be attributed to this work.

But I did had a lot of fun, trying to get different features out of text. Something, that I am sure will help me with my cognitive science master thesis. But even then, I did not ended up testing all the hypothesis. I did nothing with emotions, personality or topic prediction.

I also did not check the nouns hypothesis. There is a connection between genders in language (probably, I did not managed to find the scientific article about it). In Slovenian and Slovak, the negative words are usually female. So I figured that maybe the translation of nouns and then checking the gender in different language would also show something.

But other than that, I think I tested most of the ideas, that I got.

The jupyter book can be found here: Kaggle Competition

I hope that friend of mine is not too disappointed with my work. I learned a lot and I passed the exam, so I am satisfied with what I got out of it. Not sure about him. Since I got another link from him about text analysis, it could go both way. Either as a push in the last days to actually do something, or as here may be something that would be interesting to you.

Segregation of People - Python Meetup Presentation

Two days ago (on Tuesday, 19th of February) I had my first long presentation at the Python Meetup. I had the talk about segregation of people into like minded groups.

I have to admit, I was so nervous, that I skipped most of the programming part, and I only did mostly the cognitive science stuff. Thinking back, I can sort of understand it. I am afraid to present in front of a more knowledgeable audience. So, somebody that knows more than me. So I skipped the parts, where they would know more than me.

This is the presentation: https://github.com/sarajaksa/DataAnalysis/blob/master/Segregation%20of%20People%20into%20Like-Minded%20Groups%20[Python%20MeetUp%20Ljubljana%20February%202019].ipynb

But now the ice is broken. I still think, that if I am to get used to speaking in front of the audience like that, the lightning talks are going to be a lot better choice. I already have some ideas for it (which are mostly projects, that I want to do or I will do or I am doing right now).

There is no video available this time. Thankfully. I can still not watch myself at all.

Get Nearby Places with Open Street Maps and Overpass API

Well, one of the things, that I will have to implement in my app in the near future was the suggestion for nearby places. The person showing my this recommended Foursquare API, but I decided to first try something a bit more open source.

Open Street Map has some great maps, so this was my first try. Would it be possible to get what I wanted with it? They ask not to get too much data in a day. My test query produced a little less than 5KB, so I hope that would not be much. In case we ever get a lot of users, I am sure, we could host a subset of the data on our own servers.

I have to admit, it took me a lot more time to find out what API to use, and to try and research the difference between QL and XML and why there are two languages and how can each be used, that actually trying to understand how to make queries.

This is quite different from school work. They well you, where the data is and it one form, and then they make you master querying. Or maybe they had a good enough effect.

Well, before I tried to implement it inside of react-native, I tried it in a python code. The below code queries the places near faculty rooms for my study program. They only list the names entries, so they might be some missing. One can chance the latitude, longitude and range easily.

from bs4 import BeautifulSoup
import requests

lat = 46.04908
lon = 14.50468
rad = 50

data = requests.get("http://overpass-api.de/api/interpreter?data=<query type='node'><around lat='" + 
                    str(lat) + 
                    "' lon='" + 
                    str(lon) + 
                    "' radius='" + 
                    str(rad) + 
                    "'/></query><print/>")

data = data.text
data=BeautifulSoup(data, features="xml")
all_nearby_locations = data.find_all("tag", {"k":"name"})
[loc["v"] for loc in all_nearby_locations]

I don't know, bit for now, it seems like a relatively good solution. Now to figure out, how I am going to get these names out of the XML in react-native (hopefully without additional libraries - I have the feeling I use too many of them already).