Table 1.
List of patents registered by CPC, 2017 [28].
Fig 1.
Doc2vec models: DM model, Dbow model [14].
Fig 2.
Research framework: The framework of this study has three steps.
The first step is to collect data and extract the SAO structure, which will be used in future steps. And the second step is to extract words and sentence vectors using doc2vec algorithm and update the sentence vector. In the final step, SAO vectors and sentence vectors are obtained.
Fig 3.
Example of SAO extraction.
Fig 4.
Word and sentence vector embedding and updating.
Fig 5.
An algorithm for determining activeness of each document.
Table 2.
Collect data.
Table 3.
Example of prepared learning data in Iot field.
Table 4.
Example of SAO structure extracted by the learning data.
Fig 6.
Visualization of word vector ‘technique’.
Table 5.
Top 5 words similar with “technique” and “transmission” using Doc2vec result.
Table 6.
Example of the average similarity between sentence vector and word vector.
Fig 7.
Visualization of the 5285th sentence vector.
Fig 8.
Visualization of the 5285th FE sentence vector.
Table 7.
Example of the labeled SAO structure: Label 32, 93.
Table 8.
Example of SAO vector: Sentence 1134, 2261, 206 in patent 218.
Table 9.
Uses of an SAO vector with the same label.
Table 10.
Example of document vector: Patent 1, 2, 3.
Table 11.
Example of document similarity matrix using SAO frequency.
Table 12.
Example of document similarity matrix using Doc2Vec.
Table 13.
Example of document similarity matrix using SAO2Vec.
Table 14.
Patent clustering using spectral clustering.