---------- Forwarded message --------- From: Naren Ramakrishnan Date: Sat, Nov 12, 2022 at 8:05 AM Subject: Re: update To: Mares, David Cc: Brian Mayer Thank you very much David - very very useful! Reg your comment that we are looking for primarily surges of migration - that is true, in that ODNI would like to be ready for a sudden surprise but they are also interested in the total count(s) of people - but we will be able add a default baseline value to a surge-predicted value, to get this total count. We do have multiple news sources and we will get to separating them into country of origin (I am sure we likely have that information) so that as you say we can separate out the context in which specific keywords are used. This will be done using our geolocator. For now we want to use keywords to net a broad collection of articles and then we can distinguish the contextual information in which different keywords are used. So we can think of this stage as a very broad/weak signal just to make sure we don't lose anything that could be relevant later on. We will add these words to our list - please do let us know if there are other push/pull - and also transit issues as you mentioned. Thanks a lot! Naren On Sat, Nov 12, 2022 at 12:54 AM Mares, David wrote:

Naren and Brian,

I've been reading and conceptualizing the task and reviewing the current spreadsheet.

We are focusing on *sudden forced migration*, or *migration **surges * *NOT* on migration per se or long-term migration dynamics. We can think in terms of ‘triggers’, as we did with civil unrest.

The ‘triggers’ are highly likely to be context-dependent and if we only have one data source (major newspapers in English and Spanish?) that could overwhelm the specific context and give us no correlation – e.g., in Costa Rica a proliferation of the words “Cubans”, “Venezuelans” “Nicaraguans” or “Haitians” is likely to refer to the large influx of refugees and irregular migrants so that would be useful to track. But if we can’t separate out the Costa Rican context, then any payoff to us will be overwhelmed by the general use of these words in non-Costa Rican related news stories.

The words in the attached file are my initial take. A few of them are already included in the current spreadsheet, but most are not. Before pursuing this line of work, I thought it would be best to get your feedback.

Thanks,

David ------------------------------ *From:* Naren Ramakrishnan *Sent:* Wednesday, November 9, 2022 1:53 PM *To:* Mares, David ; Brian Mayer *Subject:* Notes from today's call

Hi David,

So great catching up with you today! Below are the notes from the call.

Brian, can you share the current spreadsheet with David?

Best, Naren

- proportion of minor, timing, origin/destination - push/pull factors, transit issues, blockages - lots of literature: sending and receiving communities - less chaos in these flows than we think, lots more organization and communication. These are high risk movements, they don’t do it willy nilly. - various types of triggers: Salvadorian govt gets into a confusion with US government, so it doesn’t do as much as it has done to curb migrants - keywords: not only in push and pull factors but also in transit arena. - pull factors: jobs, job opportunities, benefits (does not payout as a big attraction) eg CA would permit DACA students to remain in the University system, community ties (people in communities that have ties back to communities in Mexico/Nicaragua), community contacts that help people get a job.. - push factors: convince people to undertake the voyage that has associated range of costs, local violence (targeted violence toward particular communities, e.g., drug gangs shooting not just mayor but also school principal), economic collapse, natural resource issues

-- Naren Ramakrishnan http://people.cs.vt.edu/~naren/ Thomas L. Phillips Professor of Engineering Director, Sanghani Center for AI and Data Analytics https://sanghani.cs.vt.edu Director, NSF NRT UrbComp https://sanghani.cs.vt.edu/academics/urban-computing/ Department of Computer Science, Virginia Tech On Twitter as @profnaren -- Naren Ramakrishnan http://people.cs.vt.edu/~naren/ Thomas L. Phillips Professor of Engineering Director, Sanghani Center for AI and Data Analytics https://sanghani.cs.vt.edu Director, NSF NRT UrbComp https://sanghani.cs.vt.edu/academics/urban-computing/ Department of Computer Science, Virginia Tech On Twitter as @profnaren