Enriching website content with extracted feature multi-dimensional vector comparison

A method for enriching contents of a current website, the method comprising:

• obtaining a corpus from the current website and other websites, and extracting object features from the corpus, wherein the corpus comprises specifications of the object and user reviews about the object;
• according to the corpus, constructing multi-dimensional vectors for the extracted features;
• for a specific feature, making similarity comparison of its multi-dimensional vectors and multi-dimensional vectors of other extracted features;
• and
• determining features with similarities being higher than a predetermined threshold as a same feature, and reinforcing the current website with at least one feature which is different from those on the website and corresponding attributes of the at least one feature, wherein the similarities between the multi-dimensional vectors of various features are calculated as cosine similarities:
• sin
• ⁡
• (
• v
• i
• ,
• v
• j
• )
• =
• cos
• ⁢
• ⁢
• θ
• =
• ∑
• k
• =
• 1
• k
• =
• n
• ⁢
• ⁢
• w
• ⁡
• (
• f
• k
• ⁡
• (
• v
• i
• )
• )
• ×
• w
• ⁡
• (
• f
• k
• ⁡
• (
• v
• j
• )
• )
• (
• ∑
• k
• =
• 1
• k
• =
• n
• ⁢
• ⁢
• w
• 2
• ⁡
• (
• f
• k
• ⁡
• (
• v
• i
• )
• )
• )
• ⁢
• (
• ∑
• k
• =
• 1
• k
• =
• n
• ⁢
• ⁢
• w
• 2
• ⁡
• (
• f
• k
• ⁡
• (
• v
• j
• )
• )
• )
• wherein,
• vi represents a multi-dimensional vector of the i-th feature, vj represents a multi-dimensional vector of the j-th feature;
• w(fk(vi)) represents the weight of the k-th dimension of the multi-dimensional vector vi, and
• w(fk(vj)) represents the weight of the k-th dimension of the multi-dimensional vector vj.

A method for enriching contents of a website, the method comprising:

• obtaining a corpus from the website and other websites, and extracting object features from the corpus, wherein the corpus comprises specifications of the object and user reviews about the object;
• according to the corpus, constructing multi-dimensional vectors for the extracted features;
• for a specific feature, making similarity comparison of its multi-dimensional vectors and multi-dimensional vectors of other extracted features;
• and
• determining features with similarities being higher than a predetermined threshold as a same feature, and reinforcing the website with at least one feature which is different from those on the website and corresponding attributes of the at least one feature, wherein comparing the similarity of multi-dimensional vectors of the specific feature and multi-dimensional vectors of other extracted features comprises:
• aligning dimensions of multi-dimensional vectors of all extracted features;
• calculating mutual information between the specific feature in the corpus and each dimension of its multi-dimensional vectors as a weight of each dimension;
• and
• calculating similarities between multi-dimensional vectors of various features according to the weights, and based on cosine similarities.