f2f recap and discussion on discovery

<kaz> scribenick: DanhLePhuoc

<kaz> Dusseldorf f2f agenda

<kaz> Dusseldolf f2f minutes

we're currently have 3 tracks in LD task forces but only the outcomes of two tracks were presented at the F2F

there are two streams of on vocabularies

the first one is on advocating in bridging ontologies

<DarkoAnicic> iot.schema.org : http://www.w3.org/2017/07/wot-f2f/slides/iotschema-wotf2f.pdf

and another one is leaning towards building by iot.schema.org leading by michael koster

there're several separated efforts in annotating semantics/ontologies to TD

but quite some of them have some difficulties in dealing with ontologies and SW technologies and lack of devices during F2F

right now there's no use of TD repository because there's no user interface to query/discover the repository to find usable things

<inserted> scribenick: mjkoster

danh: provide a json interface for queries?
... basic vocabulary to discover things

darko: doing semantic queries was the difficulty

mkoster: maybe we need a template based discovery system where you fill out a TD type template to express what you need
... looks like part of a recipe

darko: may be an approximate match using a recipe to compose a possible solution from different ingredients
... servients register TDs to the repository and can be discovered "bottom-up" but it doesnt work very well
... top-down may be a better solution
... using recipes
... servient to servient

danh: who will create the recipes?

darko: created by humans who want to create an application

danh: will we put this into the best practices?

darko: we should look at use cases for discovery and decide what the better pattern is
... top-down or bottom up... the current pattern is not working because it is difficult for users to form sparql queries

danh: turn recipes into sparql queries or make simple js language

darko: what kind of discovery API is needed?

kaz: what is the target? local area? could use mDNS/bonjour on a local network
... do we need some new 21st century DNS for web of things?

danh: we also need to think about discovery above the network layer

kaz: is it the assumption that TD repository contains all of the addresses?

danh: yes, everything will be in TD

kaz: there may be some use for network discovery, finding a link

<kaz> https://www.w3.org/TR/2017/NOTE-mmi-mc-discovery-20170202/

<inserted> scribenick: kaz

kaz: think there are several states, e.g., IDLE, AVAILABLE, and we need to think about how to handle those states using management thing api or LD recipe
... from my viewpoint, "TD Recipe" is kind of high-level abstraction of Management Thing APIs.

<inserted> scribenick: DanhLePhuoc

DarkoAnicic: which task forces are managing the discovery requirements?

TD TF is discussing a bit on that https://w3c.github.io/wot/current-practices/wot-practices.html#td-discovery

https://github.com/thingweb/thingweb-repository/

<inserted> scribenick: kaz

kaz: Darko, are you planning to collaborate with TD guys and Management Thing guys?

darko: yes, would like to generate some proposal and bring that to the main call, e.g., maybe the week after.

[ adjourned ]