NETWORK Procedure

specifies whether to calculate Adamic-Adar node similarity. You can specify the following values:

By default, ADAMICADAR=FALSE.

specifies the maximum number of lowest-ranked similarity pairs to output in the OUTSIMILARITY= table, where number can be any integer greater than or equal to 0. If more than one similarity measure is requested, you must use the ORDERBY= option to specify which similarity measure to use for ranking.

specifies whether to calculate common neighbors node similarity. You can specify the following values:

By default, COMMONNEIGHBORS=FALSE.

specifies the convergence threshold for vector similarity. If the convergence value drops below this threshold before the procedure completes the number of samples that are requested in the NSAMPLES= option, the vector embeddings training phase terminates early. By default, CONVERGENCETHRESHOLD=0.01.

specifies whether to calculate cosine node similarity. You can specify the following values:

By default, COSINE=FALSE.

specifies whether to calculate Dice node similarity. You can specify the following values:

By default, DICE=FALSE.

specifies whether to calculate vector embeddings. You can specify the following values:

By default, EMBED=FALSE when VECTOR=FALSE, and EMBED=TRUE when VECTOR=TRUE.

specifies the names of one or more data variables to be used as precalculated vector embeddings. The value of the variables can be numeric only. Each variable must exist in the input table that is specified in the NODES= option, and a nonmissing value must exist for each variable for each node that appears in the graph. This option is ignored unless VECTOR=TRUE or EMBED=TRUE is specified.

specifies whether to calculate Jaccard node similarity. You can specify the following values:

By default, JACCARD=TRUE.

specifies the maximum similarity score to output in the OUTSIMILARITY= table, where number must be in the range . The MAXSCORE= option is supported only for the Jaccard and vector similarity measures; it does not apply to the Adamic-Adar, common neighbors, cosine, or Dice similarity measure. If you specify both JACCARD=TRUE and VECTOR=TRUE, node pairs are output if either the Jaccard or vector similarity score satisfies the maximum similarity condition. By default, MAXSCORE=1.

specifies the minimum similarity score to output in the OUTSIMILARITY= table, where number must be in the range . The MINSCORE= option is supported only for the Jaccard and vector similarity measures; it does not apply to the Adamic-Adar, common neighbors, cosine, or Dice similarity measure. If you specify both JACCARD=TRUE and VECTOR=TRUE, node pairs are output if either the Jaccard or vector similarity score satisfies the minimum similarity condition. By default, MINSCORE=0.

specifies the number of dimensions for node-embedding vectors. By default, NDIMENSIONS=100.

specifies a multiplier for the number of negative training samples per positive training sample for the vector algorithm. If NSAMPLES=s, the vector algorithm uses s positive samples and number times s negative samples. By default, NEGATIVESAMPLEFACTOR=5.

specifies the number of training samples for the vector algorithm. By default, NSAMPLES=, where is the number of links in the input graph. For distributed execution, this number of training samples is processed per machine.

specifies the similarity measure to use for ranking the similarity pairs to output in the OUTSIMILARITY= table. This option is ignored if neither the TOPK= option nor the BOTTOMK= option is specified.

specifies the output data table to contain the convergence curves for vector similarity. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the section Using CAS Sessions and CAS Engine Librefs in Chapter 1, Introduction.

specifies the output data table to contain the similarity scores between pairs of nodes. libref.data-table is a two-level name, where libref refers to the library, and data-table specifies the name of the output data table. For more information about this two-level name, see the section Using CAS Sessions and CAS Engine Librefs in Chapter 1, Introduction.

specifies the type of proximity to use in the vector algorithm. You can specify the following values:

By default, PROXIMITYORDER=SECOND.

specifies the sink node for node similarity calculations.

specifies the source node for node similarity calculations.

specifies whether to output sparse similarity data. You can specify the following values:

You cannot specify SPARSE=TRUE when you specify VECTOR=TRUE. By default, SPARSE=FALSE.

specifies the maximum number of highest-ranked similarity pairs to output in the OUTSIMILARITY= table, where number can be any integer greater than or equal to 0. If more than one similarity measure is requested, you must use the ORDERBY= option to specify which similarity measure to use for ranking.

specifies whether to calculate vector node similarity. You can specify the following values:

When you specify VECTOR=TRUE, you cannot specify SPARSE=TRUE. By default, VECTOR=FALSE.