Term suggester : strange results

Elastic Stack Elasticsearch
1

Hi, everyone,

I'm trying to figure put some discrepencies (I think) in the results of my
suggesters, with ES V0.90.5.

My indices are big and can contain a wide array of language.
when I do this (NB : tomate is the french for tomato)
;

{
"query": {
"match_all": {}
},
"suggest": {
"my_suggester": {
"text": "tomate",
"term": {
"field": "my_field"
}
}
}
}

Es doesn't suggest "tomato". And the terms it suggest are rather
low-scoring (it even suggests "tote", with a measly 0.5 score)

however, if I do this :

{
"query": {
"match_all": {}
},
"suggest": {
"my_suggester": {
"text": "tomato",
"term": {
"field": "my_field"
}
}
}
}

Es suggests "tomate", along with a buch of lower scoring terms.

As far as I undestand, tomate is as close to tomato as tomato is close to
tomate, and thus ES should suggest tomato when I'm asking for tomate.
I'm positive that the two terms are present in the indices; and brought
back by my request, so that would not be the issue.

So, I wonder ..
Is there something I did not understand regarding suggesters?
Is that behaviour normal?
Is it due to the older version of ES I'm still using?

If anyone using suggesters would help me make sense of this, that'd be
helpful.
Thanks

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/d4beac46-8520-472d-9acc-fdaa02dd5d4e%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

2

On Thu, Dec 4, 2014 at 7:27 AM, DH ciddp195@gmail.com wrote:

Hi, everyone,

I'm trying to figure put some discrepencies (I think) in the results of my
suggesters, with ES V0.90.5.

My indices are big and can contain a wide array of language.
when I do this (NB : tomate is the french for tomato)
;

{
"query": {
"match_all": {}
},
"suggest": {
"my_suggester": {
"text": "tomate",
"term": {
"field": "my_field"
}
}
}
}

Es doesn't suggest "tomato". And the terms it suggest are rather
low-scoring (it even suggests "tote", with a measly 0.5 score)

however, if I do this :

{
"query": {
"match_all": {}
},
"suggest": {
"my_suggester": {
"text": "tomato",
"term": {
"field": "my_field"
}
}
}
}

Es suggests "tomate", along with a buch of lower scoring terms.

As far as I undestand, tomate is as close to tomato as tomato is close to
tomate, and thus ES should suggest tomato when I'm asking for tomate.
I'm positive that the two terms are present in the indices; and brought
back by my request, so that would not be the issue.

So, I wonder ..
Is there something I did not understand regarding suggesters?
Is that behaviour normal?
Is it due to the older version of ES I'm still using?

If anyone using suggesters would help me make sense of this, that'd be
helpful.
Thanks

Try setting these:
"suggest_mode": "always",
"size": 100000,
"max_term_freq": 0.65

and then backing off from there. Your suggestion should come back with all
the settings and you'll learn more about what is going on.

Depending on how the documents are sharded you can see changes in
behavior. Its more likely when you have fewer documents.

Nik

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/CAPmjWd0rdwOoSBsSdVMuq_P7ntqn%2B7YOOZpD_PyFcqg4%3D99vbA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

3

Yes ! Thank a lot, "mode : always" did the trick.

Now, I just have some strange frequency numbers .. suggesters boasts 203
"tomato", whereas a query only return 97.
The frequencies are less important, so I guess I'll be able to live with
that.

Thank again

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/8e58d7a1-5cac-4535-afb9-fe58be574aa5%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.