search

mchozhang / SinsOnTwitter

4 stars 3 forks source link

Sin Topic #12

Open CacarianLong opened 5 years ago

CacarianLong commented 5 years ago

26 AURIN Data

CacarianLong commented 5 years ago

Visualize idea:

CacarianLong commented 5 years ago
CacarianLong commented 5 years ago

1. Road Rage Application is a web application that shows the extent of road rage in different areas of Victoria in real time. By analysing tweets related to road rage and the real-time traffic data, it predicts the risk of car crash in an area based on map. Functions:

  1. Harvest tweets in real time (every two mins) based on Victoria. Extract fields of tweet texts, geographical info, time tag.
  2. sentiment analyse the tweets. Classify a tweet as road rage shown or not using sentiment analysis. Calculate the percentage of road rage tweets in each area, e.g., a grid cell.
  3. Monitor traffic in Victoria in real time by streaming Vicroads Open Data API. Visualise the traffic jam based on map.
  4. Predict the chance of car crash or risk in an area based on the information above, using regression analysis.
  5. Users can zoom in or out the map to check the extent of road rage in different areas.

API link: [1] real time traffic data https://vicroadsopendata-vicroadsmaps.opendata.arcgis.com/datasets/bluetooth-travel-time-updates-every-2-minutes/geoservice?geometry=-281.25%2C-52.268%2C281.25%2C52.268 image

Other data may be useful: [1] AURIN data of traffic crime: Fatal Crashes: https://data.aurin.org.au/dataset?sort=score+desc%2C+metadata_modified+desc&organization=vic_govt_vicroads&ext_prev_extent=94.5703125%2C-54.36775852406839%2C171.9140625%2C5.266007882805498&ext_bbox=&q=crash [2] Number of Offences in VIC (including transport regulation offences) https://data.aurin.org.au/dataset?sort=score+desc%2C+metadata_modified+desc&ext_prev_extent=94.5703125%2C-54.36775852406839%2C171.9140625%2C5.266007882805498&ext_bbox=&q=crime&organization =vic_govt_det

CacarianLong commented 5 years ago

Idea 2. Abuse terms, bulling and violence

  1. analyze the tweets data to extract {abuse wordlist, geograpyical information}
  2. calculate the frequency of each abuse term and order the frequency of each area in Victoria.
  3. Visualise the data based on map to the user according to user's query.
  4. Visualize the violence crime based on map in VIC to the user.
  5. association between result of 3 & 4 .. https://github.com/topics/profanity
relientm96 commented 5 years ago

Sin of Sloth. Sloth defined as: a habitual lack of enthusiasm for effort, or laziness We analyze tweets to check for sloth properties. Sloth properties can be seen through tweets with:

We can validate our data with AURIN (Or other official datasets) through different topics such as exercise Data: a. number of gyms/sports facilities in a grid/location https://data.aurin.org.au/dataset/vic-govt-dhhs-vic-sport-and-recreation-2015-na b. number of people with composite risk factors (smokers, alcoholics, obesity and no/low exercise rates,) https://data.aurin.org.au/dataset/tua-phidu-phidu-estimates-composite-risk-phn-2014-15-na

Advantages: Easy to collect from twitter Disadvantages: Hard to "validate" laziness from official data. (No direct relationship to laziness)

relientm96 commented 5 years ago

Sin of Greed and Envy. Envy is the desire for others' traits, status, abilities, or situation. Greed is the desire for material wealth or gain, ignoring the realm of the spiritual.

We can obtain greed information from twitter by:

Official data sets that we can use to validate envy/greed:

Advantages:

relientm96 commented 5 years ago

Sin of Wrath Anger is manifested in the individual who spurns love and opts instead for fury.

We can get twitter data on wrath with: ( a lot people rage on twitter so this is the easiest)

Official data sets:

Advantages:

Arufu commented 5 years ago

If we want to do something like all sins, we may correlate the frequency of all sin-related words with income, to see what part (in terms of income range) of the populace commit more sins: the rich? the poor? or the middle class?