WDQS - graph splitting options

	A	B	C	D	E	F	G	H	I	J	K	L	M	N
1	What to split out?	What's the size of the split-out graph in number of triples?	What's the size of the split-out graph in number of Items?	% count of all queries	Is it a semanticly sensible split?	Are we minimizing the need to query the graphs together?	How big is the affected group?	Does this work together with the plans for splitting the dumps?	other thoughts	Next Action	Who	when	How	Key Result: Success Metrics
2	scholarly articles	XXL (it's the largest semantic subgraph we have)	XXL (it's the largest semantic subgraph we have)	2.6	yes	yes from the main graph to the split-out graph but not the other way around (meaning people querying for scholarly articles will need to federate for basically everything)	S	yes		Testing Experiment, Extract of Current Queries	GL, David, Desiree, Lydia	Phabricator backlog- till Davids capacity becomes available to start tests	New End point, Users: re-write queries, Query returns are correct	Correct Results are returned for User Queries, Low Volume of Broken Queries, Fewer Slower Queries (Whollistic Graph Stability: Pre-Post data comparisons from this test), Impact on Tripples, Growth on Federated instance vs Graph to guide for future DR planning etc
3	astronomical objects	XL (it's the second-largest semantic subgraph we have)	XL (it's the second-largest semantic subgraph we have)	1.3	yes	same as for scholarly articles except that there are a few planets and stars people will expect in the main graph	XS	probably	We might have to duplicate a few planets and stars
4	humans	XL (it's the third-largest semantic subgraph we have)	XL (it's the third-largest semantic subgraph we have)	31.9	yes	no, humans are involved in many different topics people query for so they are not very contained	XXL	probably
5	Wikimedia-internal stuff (templates, categories, ...)	? probably high	? maybe low		yes	likely yes	XS	probably		check number of Items, Tripples	Lydia, Manuel, Ayesha
6	Lexemes	?	?		yes	Yes from Items to Lexemes but not from Lexemes to Items	S	yes
7	terms (labels, aliases, descriptions)	1/7th of all triples apparently for descriptions	-		maybe	very likely		probably	Would maybe not be a proper split in our sense of the word	check numbers for labels, aliases to make this compelling. Decision: to integrate into SPARQL query engine is a big architectural change- not sure to do this	Lydia, Nikki
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100