Domain name classification systems and methods
Disclosed is a domain engineering analysis solution that determines relevance of a domain name to a brand name in which a domain name, brand name, and identification of a substring of the domain name may be provided to or obtained by a computer embodying a domain engineering analyzer. A list of features may be determined. The list of features may include a lexicon, or a set of key-value pairs that encode information about terms included as substrings in the domain name. Determining the features may include obtaining a language model for each term, analyzing a cluster of language models closest to the obtained language model, and determining and scoring a relevance of each term to the brand name. The determined relevance and score of each term may be provided to a client. This relevance analysis can be dynamically applied in an online process or proactively applied in an offline process.
1. A method, comprising:
determining, by a computer in pre-processing, terms that are relevant to a brand, the determining including:
analyzing words used in domain names registered to the brand;
for each term of interest, obtaining a language model and language models that are most similar to the term of interest, the obtaining including:
computing a distance between the language model for the term of interest and each of a plurality of language models associated with a set of words; and
determining a cluster of least-distant language models from the plurality of language models based on the distance between the language model for the term of interest and a respective language model of the plurality of language models associated with the set of words, the cluster of least-distant language models associated with neighbors of the term of interest;
computing a score indicating a relevance of the term of interest to the brand, the computing including determining a number of the neighbors appearing in the domain names registered to the brand;
determining, by a computer in processing an input domain name, character counts and locations of the characters in the input domain name;
for each term indicated by the character counts and the locations of the characters in the input domain name, retrieving a relevance score computed in the pre-processing;
determining a final score for the input domain name based at least on the relevance score of each term in the input domain name; and
providing the final score to a classifier for classifying the input domain name in relevance to the brand.
2. The method of claim 1 , wherein the processing the input domain name includes obtaining a substring of the input domain name that is at least an approximate match to the brand name.
3. The method of claim 2 , wherein the substring of the input domain name includes:
an exact match to the brand name, or
an approximate match that matches the brand name within a predetermined threshold measure of distance from the brand name.
4. The method of claim 2 , wherein the obtaining the substring of the input domain name that is at least an approximate match to the brand name includes:
obtaining a pre-processed set of key-value pairs, wherein the pre-processed set of key-value pairs includes one or more index values indicating a location of the substring of the first domain name that is at least an approximate match to the brand name as a value of a first pre-processed key-value pair of the pre-processed set of key-value pairs, as the substring of the input domain name.
5. The method of claim 1 , further comprising: obtaining the brand name.
6. The method of claim 1 , further comprising: obtaining a pre-processed set of key-value pairs that includes the input domain name as a value of a first pre-processed key-value pair, and the brand name as a value of a second pre-processed key-value pair.
7. The method of claim 1 , wherein the obtaining includes analyzing a plurality of word vectors closest to a word vector for the term of interest.
8. The method of claim 1 , wherein the determining the duster of least-distant language models includes analyzing a plurality of language models within a predetermined threshold distance from the language model for the term of interest.
9. A system, comprising:
a processor;
a non-transitory computer-readable medium; and
stored instructions translatable by the processor for:
determining, in pre-processing, terms that are relevant to a brand, the determining including:
analyzing words used in domain names registered to the brand;
for each term of interest, obtaining a language model and language models that are most similar to the term of interest, the obtaining including:
computing a distance between the language model for the term of interest and each of a plurality of language models associated with a set of words; and
determining a cluster of least-distant language models from the plurality of language models based on the distance between the language model for the term of interest and a respective language model of the plurality of language models associated with the set of words, the cluster of least-distant language models associated with neighbors of the term of interest;
computing a score indicating a relevance of the term of interest to the brand, the computing including determining a number of the neighbors appearing in the domain names registered to the brand;
determining, in processing an input domain name, character counts and locations of the characters in the input domain name;
for each term indicated by the character counts and the locations of the characters in the input domain name, retrieving a relevance score computed in the pre-processing;
determining a final score for the input domain name based at least on the relevance score of each term in the input domain name; and
providing the final score to a classifier for classifying the input domain name in relevance to the brand.
10. The system of claim 9 , wherein the processing the input domain name includes obtaining a substring of the input domain name that includes:
an exact match to the brand name, or
an approximate match that matches the brand name within a predetermined threshold measure of distance from the brand name.
11. The system of claim 10 , wherein the obtaining includes:
obtaining a pre-processed set of key-value pairs that includes the input domain name as a value of a first pre-processed key-value pair, and the brand name as a value of a second pre-processed key-value pair of the pre-processed set of key-value pairs.
12. The system of claim 10 , wherein the obtaining includes:
obtaining a pre-processed set of key-value pairs that includes a substring of the input domain name that is at least an approximate match to the brand name as a value of a first pre-processed key-value pair of the pre-processed set of key-value pairs.
13. The system of claim 12 , wherein the pre-processed set of key-value pairs includes one or more index values indicating a location of the substring of the input domain name that is at least an approximate match to the brand name.
14. The system of claim 12 , wherein the pre-processed set of key-value pairs includes one or more index values indicating a starting location and an ending location of the substring of the input domain name that is at least an approximate match to the brand name.
15. The system of claim 9 , further comprising:
obtaining the brand name.
16. The system of claim 9 , wherein the obtaining includes analyzing a plurality of word vectors closest to a word vector for the term of interest.
17. The system of claim 9 , wherein the determining the cluster of least-distant language models includes analyzing a plurality of language models within a predetermined threshold distance from the language model for the term of interest.
18. A computer program product having a non-transitory computer-readable medium storing instructions translatable by the processor for:
determining, in pre-processing, terms that are relevant to a brand, the determining including:
analyzing words used in domain names registered to the brand;
for each term of interest, obtaining a language model and language models that are most similar to the term of interest, the obtaining including:
computing a distance between the language model for the term of interest and each of a plurality of language models associated with a set of words; and
determining a cluster of least-distant language models from the plurality of language models based on the distance between the language model for the term of interest and a respective language model of the plurality of language models associated with the set of words, the cluster of least-distant language models associated with neighbors of the term of interest;
computing a score indicating a relevance of the term of interest to the brand, the computing including determining a number of the neighbors appearing in the domain names registered to the brand;
determining, in processing an input domain name, character counts and locations of the characters in the input domain name;
for each term indicated by the character counts and the locations of the characters in the input domain name, retrieving a relevance score computed in the pre-processing;
determining a final score for the input domain name based at least on the relevance score of each term in the input domain name; and
providing the final score to a classifier for classifying the input domain name in relevance to the brand.
19. The computer program product of claim 18 , wherein the processing the input domain name includes obtaining a substring of the input domain name that is at least an approximate match to the brand name.
20. The computer program product of claim 19 , wherein the substring of the input domain name includes:
an exact match to the brand name, or
an approximate match that matches the brand name within a predetermined threshold measure of distance from the brand name.