Provide a point-view for the geo-location tagged tweets

[JavierJia]

plot the tweets by it's geolocation as a point.

It could be implemented by using the geocell function of the new data model.

[JavierJia]

The aggregation result is easy to show. The harder problem is to show the information on data points. The details Query could be implemented as group by the geocell function and get the "topK" most recent tweet within each group.

[JavierJia]

The "topK" may not serve the purpose directly. What we need is a "top10%" instead of "top100" result. Does *db support this query?

[JavierJia]

@simonmssu

[JavierJia]

Get an idea from a discussion in SIGSPATIAL: issue an aggregation query for each very small cell; zoom-in can be done with another query with even smaller cells, etc.

[JavierJia]

@zonghengma we can do it step by step. The first step just show all points within a small region. e.g., when we zoom to a layer finner then city, we show all points in that region.

One problem is that the sample data may not have many point data.

[JavierJia]

@HotLemonJuice , could you move the issue #274 content you just created to here? thanks!

[ShengjieXu]

So far we are using a heat map to show the Tweets on the map. It is a high-level description of the data.

Now if we can show the exact location of each Tweet, we can make the connection between the data and the geolocation closer.

A good example by MapD: https://www.mapd.com/demos/tweetmap/

However, since there are too much data to load in the browser, we need use some tricks to avoid crashing the browser. We can start implementing this feature in small areas of the map, e.g. city level.

[ShengjieXu]

There aren't much point data in the sample data. How do I add fake data into the DB?

[JavierJia]

we can generate a sample data. for now, you may use the field of "bounding_box": rectangle("-74.146932,40.643773 -74.0658,40.697794") }? you can just pick the first value, and add some random value.

[JavierJia]

you can take a look at the sample.data under script folder. Here is an example of the place field

"place": { "country": "United States", "country_code": "United States", "full_name": "Houston, TX", 
"id": "1c69a67ad480e1b1", "name": "Houston", "place_type": "city", "bounding_box": 
rectangle("-95.823268,29.522325 -95.069705,30.154665") }

[ShengjieXu]

Notes for future dev:

• use city ID and existing coordinate to show points (if some piece of data doesn't contain them, we just don't show them)
• use L.circleMarker to display points on the map
• checkout the js controller for the hashtag. It has the information of city ID...

Now, I'm switching off to Normalization as it is more urgent