You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by Borbála Siklósi <si...@gmail.com> on 2010/10/21 11:20:04 UTC
mean shift clustering result
I am trying to run mean-shift clustering on a set of documents. I have done
the vectors and ran the algorithm but I cannot resolve the results. I
applied cluster dump as well. My result file looks like this:
MSV-6782{n=3 c=[antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241] r=[]}
Top Terms:
nevirapine => 8.59228229522705
hiv => 7.710737705230713
ne => 7.339519023895264
vye => 6.376708507537842
antiretroviral => 5.604918003082275
peen => 5.501239776611328
ra => 5.421196937561035
others => 4.55509614944458
cure => 4.385105133056641
medicines => 4.224946022033691
Weight: Point:
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
Where n=1 I have the document url and vector at r=...
How can I see the clusters and the documents belonging to each cluster?
RE: mean shift clustering result
Posted by Jeff Eastman <je...@Narus.com>.
I suggest wrapping your input vectors in NamedVector wrappers so that the document name will flow through the clustering algorithm and be visible in the cluster dumper output. I think seq2sparse does this now, are you running on trunk?
-----Original Message-----
From: Borbála Siklósi [mailto:siklosib@gmail.com]
Sent: Thursday, October 21, 2010 2:20 AM
To: user@mahout.apache.org
Subject: mean shift clustering result
I am trying to run mean-shift clustering on a set of documents. I have done
the vectors and ran the algorithm but I cannot resolve the results. I
applied cluster dump as well. My result file looks like this:
MSV-6782{n=3 c=[antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241] r=[]}
Top Terms:
nevirapine => 8.59228229522705
hiv => 7.710737705230713
ne => 7.339519023895264
vye => 6.376708507537842
antiretroviral => 5.604918003082275
peen => 5.501239776611328
ra => 5.421196937561035
others => 4.55509614944458
cure => 4.385105133056641
medicines => 4.224946022033691
Weight: Point:
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
1.0: [antiretroviral:5.605, ask:1.315, care:1.343, cure:4.385,
have:1.116, health:1.331, hiv:7.711, may:1.120, medicine:2.628,
medicines:4.225, ne:7.340, nevirapine:8.592, other:1.940, others:4.555,
peen:5.501, pharmacist:1.320, provider:1.351, purpose:1.364, purposes:1.365,
question:1.363, questions:1.364, ra:5.421, spread:4.175, stop:3.400,
treat:1.323, use:1.467, used:1.501, vye:6.377, you:1.716, your:1.241]
Where n=1 I have the document url and vector at r=...
How can I see the clusters and the documents belonging to each cluster?