Syntactically and/or semantically relevant alternative domain name generation

1. A computer-implemented method, comprising:

accessing a domain name via a processor of a computing device configured to generate domain names;

determining, via the processor, at least one word in the domain name;

determining, via the processor, at least one alternative word for the at least one word in the domain name;

determining, via the processor, whether the at least one alternative word is contextually relevant by:

calculating a probability of contextual relevance of the at least one alternative word against a language model, wherein the language model is generated by a shallow feed-forward neural network, wherein the shallow feed-forward neural network calculates the probability of the contextual relevance based on a plurality of vectors associated with training words used to train the shallow feed-forward neural network, wherein the training words are in a same language as the at least one word in the domain name; and

selecting the at least one alternative word if the probability is above a predetermined threshold;

generating, via the processor, at least one alternative domain name based on a determination that the at least one alternative word is contextually relevant and a determination that the at least one alternative domain name is available for registration; and

providing, via a user interface, the at least one alternative domain name to a user for selection by the user for registration.

2. The computer-implemented method of claim 1 , wherein determining at least one word in the domain name includes:

determining at least two words in the domain name; and

determining whether the domain name includes a multiword expression.

3. The computer-implemented method of claim 1 , further comprising:

ranking the at least one alternative domain name; and

wherein providing the at least one alternative domain name includes:

providing the at least one alternative domain name to the user in a ranked order based on the ranking.

4. The computer-implemented method of claim 1 , wherein the training words include words from zone data of a zone file.

5. The computer-implemented method of claim 1 , wherein the shallow feed-forward neural network calculates the probability of the contextual relevance based on a clustering technique applied to the plurality of vectors associated with the training words used to train the shallow feed-forward neural network,

wherein the plurality of vectors are calculated based on a continuous window model, and

wherein the clustering technique applied to the plurality of vectors associated with the training words used to train the shallow feed-forward neural network comprises a k-means clustering technique.

6. The computer-implemented method of claim 1 , wherein the training words used to train the shallow feed-forward neural network are unlabeled.

7. The computer-implemented method of claim 1 , wherein the shallow feed-forward neural network implements a hierarchical softmax technique, wherein input projections are concatenated in a projection layer of the shallow feed-forward neural network.

8. An apparatus configured to generate domain names, the apparatus comprising:

a memory storing instructions; and

at least one processor, operably connected to the memory, implemented at least in part in hardware, and configured to execute the instructions to perform operations comprising:

accessing a domain name;

determining, via the at least one processor, at least one word in the domain name;

determining, via the at least one processor, at least one alternative word for the at least one word in the domain name;

determining, via the at least one processor, whether the at least one alternative word is contextually relevant by:

calculating a probability of contextual relevance of the at least one alternative word against a language model, wherein the language model is generated by a shallow feed-forward neural network trained using data that includes zone data from a zone file, wherein the shallow feed-forward neural network calculates the probability of the contextual relevance based on a plurality of vectors associated with training words used to train the shallow feed-forward neural network, wherein the training words are in a same language as the at least one word in the domain name; and

selecting the at least one alternative word if the probability is above a predetermined threshold;

generating, via the at least one processor, at least one alternative domain name based on a determination that the at least one alternative word is contextually relevant and a determination that the at least one alternative domain name is available for registration; and

providing, via a user interface, the at least one alternative domain name to a user for selection by the user for registration.

9. The apparatus of claim 8 , wherein determining at least one word in the domain name includes:

determining at least two words in the domain name; and

determining whether the domain name includes a multiword expression.

10. The apparatus of claim 8 , the operations further comprising:

ranking the at least one alternative domain name; and

wherein providing the at least one alternative domain name includes:

providing the at least one alternative domain name to the user in a ranked order based on the ranking.

11. The apparatus of claim 8 , wherein the training words include words from zone data of a zone file.

12. A non-transitory computer-readable medium comprising computer-interpretable instructions which, when executed by at least one electronic processor of a computing device configured to generate domain names, cause the computing device to perform operations, the operations comprising:

accessing a domain name;

determining, via the at least one electronic processor, at least one word in the domain name;

determining, via the at least one electronic processor, at least one alternative word for the at least one word in the domain name;

determining, via the at least one electronic processor, whether the at least one alternative word is contextually relevant by:

calculating a probability of contextual relevance of the at least one alternative word against a language model, wherein the language model is generated by a shallow feed-forward neural network trained using data that includes zone data from a zone file, wherein the shallow feed-forward neural network calculates the probability of the contextual relevance based on a plurality of vectors associated with training words used to train the shallow feed-forward neural network, wherein the training words are in a same language as the at least one word in the domain name; and

selecting the at least one alternative word if the probability is above a predetermined threshold;

generating at least one alternative domain name based on a determination that the at least one alternative word is contextually relevant and a determination that the at least one alternative domain name is available for registration; and

providing, via a user interface, the at least one alternative domain name to a user for selection by the user for registration.

13. The computer-readable medium of claim 12 , wherein determining at least one word in the domain name includes:

determining at least two words in the domain name; and

determining whether the domain name includes a multiword expression.

14. The computer-readable medium of claim 12 , the operations further comprising:

ranking the at least one alternative domain name; and

wherein providing the at least one alternative domain name includes:

providing the at least one alternative domain name to the user in a ranked order based on the ranking.

15. The computer-readable medium of claim 12 , wherein the training words include words from zone data of a zone file.