Understanding ambiguity in a model: why this is allowed? - ssas

Until know I thought I can't have 2 paths between tables, but currently I have this in my model without any problem:
I'm creating a model for SSAS 1400 Tabular, if it matters. One more thing: I'm "coming" from PowerBI, where this is not possible to have (I can't set the relationship between Timesheet and Feladat to "both" (Timesheet = Munkaidő ráfordítások)):
And another one. Here I can't make the relationship between Megrendelés and Dolgozó-Megrendelés Reláció, because SSDT says:
Failed to save modifications to the server. Error returned: 'There are ambiguous paths between 'Feladat' and 'Dolgozó-Megrendelés Reláció': 'Feladat'->'Megrendelés'->'Dolgozó-Megrendelés Reláció' and 'Feladat'->'Timesheet'->'Dolgozó'->'Dolgozó-Megrendelés Reláció'
'.
Which I understand, but the first image is contradictory for me then...
Could someone explain these "phenomenoms" for me?

Megrendeles Rlacio ======> megrendels is not a path.
In other words, the direction symbols say you can navigate from megrendels ===> Dolgozo-Megrendeles Rlacio, not vice versa.

Related

Understanding the "Not found: Dataset ### was not found in location US" error

I know this topic has come up many times but still here I am. Data processing location seems consistent (dataset, US; query: US) and I am using backticks & long format in the FROM clause
Below are two sequences of code. The first one works perfectly:
SELECT station_id
FROM `bigquery-public-data.austin_bikeshare.bikeshare_stations`
Whereas the following returns an error message:
SELECT bikeshare_stations.station_id
FROM `bigquery-public-data.austin_bikeshare`
Not found: Dataset glassy-droplet-347618:bigquery-public-data was not found in location US
My question, thus, is why do the first lines of text work while the second doesn't?
You need to understand the different parts of the backticks:
bigquery-public-data is the name of the project;
austin_bikeshare is the name of the schema (aka dataset in BQ); and
bikeshare_stations is the name of the table/view.
Therefore, the shorter format you are looking for is: austin_bikeshare.bikeshare_stations (instead of bigquery-public-data.austin_bikeshare).
Using bigquery-public-data.austin_bikeshare means that you have a schema called bigquery-public-data that contains a table called austin_bikeshare , when this is not true.

Create subgraph query in Gremlin around single node with outgoing and incoming edges

I have a large Janusgraph database and I'd to create a subgraph centered around one node type and including incoming and outgoing nodes of specific types.
In Cypher, the query would look like this:
MATCH (a:Journal)N-[:PublishedIn]-(b:Paper{paperTitle:'My Paper Title'})<-[:AuthorOf]-(c:Author)
RETURN a,b,c
This is what I tried in Gremlin:
sg = g.V().outE('PublishedIn').subgraph('j_p_a').has('Paper','paperTitle', 'My Paper Title')
.inE('AuthorOf').subgraph('j_p_a')
.cap('j_p_a').next()
But I get a syntax error. 'AuthorOf' and 'PublishedIn' are not the only edge types ending at 'Paper' nodes.
Can someone show me how to correctly execute this query in Gremlin?
As written in your query, the outE step yields edges and the has step will check properties on those edges, following that the query processor will expect an inV not another inE. Without your data model it is hard to know exactly what you need, however, looking at the Cypher I think this is what you want.
sg = g.V().outE('PublishedIn').
subgraph('j_p_a').
inV().
has('Paper','paperTitle', 'My Paper Title').
inE('AuthorOf').
subgraph('j_p_a')
cap('j_p_a').
next()
Edited to add:
As I do not have your data I used my air-routes graph. I modeled this query on yours and used some select steps to limit the data size processed. This seems to work in my testing. Hopefully you can see the changes I made and try those in your query.
sg = g.V().outE('route').as('a').
inV().
has('code','AUS').as('b').
select('a').
subgraph('sg').
select('b').
inE('contains').
subgraph('sg').
cap('sg').
next()

Am I training my wit.ai bot correctly?

I'm trying to train my Wit.ai bot in order to recognize the first name of someone. I'm not very sure if I well understand how the NLP works so I'll give you an example.
I defined a lot of expressions like "My name is XXXX", "Everybody calls me XXXX"
In the "Understanding" table I added an entity named "contact_name" and I add almost 50 keywords like "Michel, John, Mary...".
I put the trait as "free-text" and "keywords".
I'm not sure if this process is correctly. So, I ask you:
does it matter the context like "My name is..." for the NLP? I mean...will it help the bot to predict that after this expression probably a fist name will come on?
is that right to add like 50 values to an entity or it's completly wrong?
what do you suggest as a training process in order to get the first name of someone?
You have done it right by keeping the entity's search strategy as "free-text" and "Keywords". But Adding keywords examples to the entity doesn't make any sense because a person's name is not a keyword.
So, I would recommend a training strategy which is as follows:
Create various templates of the message like, "My name is XYZ", "I am XYZ", "This is XYZ" etc. (all possible introduction messages you could think of)
Remove all keywords and expressions for the entity you created and add these two keywords:
"a b c d e f g h i j k l m n o p q r s t u v w x y z"
"XYZ" (can give any name but maintain this name same for validating the templates)
In the 'Understanding' tab enter the messages and extract the name into the entity ("contact_name" in your case) and validate them
Similarly, validate all message templates keeping the name as "XYZ"
After you have done this for all templates your bot will be able to recognise any name in a given template of the message
The logic behind this is your entity is a free-text and keyword which means it first tries to match the keyword if not matched it tries to find the word in the same position of the templates. Keeping the name same for validations helps to train the bot with the templates and learn the position where the name will be usually found.
Hope this works. I have tried this and worked for me. I am not sure how bot trains in background. I recommend you to start a new app and do this exercise.
Comment if there is any problem.
wit.ai has a pre-trained entity extraction method called wit/contact, which
Captures free text that's either the name or a clear reference to a
person, like "Paul", "Paul Smith", "my husband", "the dentist".
It works good even without any training data.
To read about the method refer to duckling.

orientdb sql update edge?

I have been messing around with orientdb sql, and I was wondering if there is a way to update an edge of a vertex, together with some data on it.
assuming I have the following data:
Vertex: Person, Room
Edge: Inside (from Person to Room)
something like:
UPDATE Persons SET phone=000000, out_Inside=(
select #rid from Rooms where room_id=5) where person_id=8
obviously, the above does not work. It throws exception:
Error: java.lang.ClassCastException: com.orientechnologies.orient.core.id.ORecordId cannot be cast to com.orientechnologies.orient.core.db.record.ridbag.ORidBag
I tried to look at the sources at github searching for a syntax for bag with 1 item,
but couldn't find any (found %, but that seems to be for serialization no for SQL).
(1) Is there any way to do that then? how do I update a connection? Is there even a way, or am I forced to create a new edge, and delete the old one?
(2) When writing this, it came to my mind that perhaps edges are not the way to go in this case. Perhaps I should use a LINK instead. I have to say i'm not sure when to use which, or what are the implications involved in using any of them. I did found this though:
https://groups.google.com/forum/#!topic/orient-database/xXlNNXHI1UE
comment 3 from the top, of Lvc#, where he says:
"The suggested way is to always create an edge for relationships"
Also, even if I should use a link, please respond to (1). I would be happy to know the answer anyway.
p.s.
In my scenario, a person can only be at one room. This will most likely not change in the future. Obviously, the edge has the advantage that in case I might want to change it (however improbable that may be), it will be very easy.
Solution (partial)
(1) The solution was simply to remove the field selection. Thanks for Lvca for pointing it out!
(2) --Still not sure--
CREATE EDGE and DELETE EDGE commands have this goal: avoid the user to fight with underlying structure.
However if you want to do it (a little "dirty"), try this one:
UPDATE Persons SET phone=000000, out_Inside=(
select from Rooms where room_id=5) where person_id=8
update EDGE Custom_Family_Of_Custom
set survey_status = '%s',
apply_source = '%s'
where #rid in (
select level1_e.#rid from (
MATCH {class: Custom, as: custom, where: (custom_uuid = '%s')}.bothE('Custom_Family_Of_Custom') {as: level1_e} .bothV('Custom') {as: level1_v, where: (custom_uuid = '%s')} return level1_e
)
)
it works well

Creating Titan indexed types for elasticsearch

I am having problems getting the elastic search indexes to work correctly with Titan Server. I currently have a local Titan/Cassandra setup using Titan Server 0.4.0 with elastic search enabled. I have a test graph 'bg' with the following properties:
Vertices have two properties, "type" and "value".
Edges have a number of other properties with names like "timestamp", "length" and so on.
I am running titan.sh with the rexster-cassandra-es.xml config, and my configuration looks like this:
storage.backend = "cassandra"
storage.hostname = "127.0.0.1"
storage.index.search.backend = "elasticsearch"
storage.index.search.directory = "db/es"
storage.index.search.client-only= "false"
storage.index.search.local-mode = "true"
This configuration is the same in the bg config in Rexter and the groovy script that loads the data.
When I load up Rexster client and type in g = rexster.getGraph("bg"), I can perform an exact search using g.V.has("type","ip_address") and get the correct vertices back. However when I run the query:
g.V.has("type",CONTAINS,"ip_")
I get the error:
Data type of key is not compatible with condition
I think this is something to do with the type "value" not being indexed. What I would like to do is make all vertex and edge attributes indexable so that I can use any of the string matching functions on them as necessary. I have already tried making an indexed key using the command
g.makeKey("type").dataType(String.class).indexed(Vertex.class).indexed("search",Vertex.class).make()
but to be honest I have no idea how this works. Can anyone help point me in the right direction with this? I am completely unfamiliar with elastic search and Titan type definitions.
Thanks,
Adam
the Wiki page Indexing Backend Overview should answer every little detail of your questions.
Cheers,
Daniel