Semantic similarity for machine learned job posting result ranking model
In an example embodiment, a machine learning algorithm is used to train a deep semantic similarity neural network to output a semantic similarity score between a candidate job search query and a candidate job search result. This semantic similarity score can then be used in a ranking phase to rank job search results in response to a first job search query.
1. A system comprising:
a non-transitory computer-readable medium having instructions stored thereon, which, when executed by a processor, cause the system to:
obtain training data pertaining to sample job posting search queries, the training data comprising sample job posting search results and indications as to which of the sample job posting search results were selected by members performing job posting search queries corresponding to the sample job posting search results;
for each of the sample job posting search queries, feed the training data corresponding to the sample job posting into a machine learning algorithm to train a deep semantic similarity neural network to output a semantic similarity score indicating semantic similarity between a candidate job search query and a candidate job search result;
identify a plurality of job search results in response to a first job search query;
for each of the plurality of job search results:
feed the job search result and the first job search query into the deep semantic similarity neural network to obtain a semantic similarity score indicating semantic similarity between the job search result and the first job search query;
rank the plurality of job search results based on the semantic similarity score for each of the job search results;
transmit the ranked plurality of job search results to a client device for display; and
augment the first job search query using terms that are semantically similar to terms in the first job search query, utilizing a semantic similarity score output from the deep semantic similarity neural network.
2. The system of claim 1 , wherein the deep semantic similarity neural network includes a hashing layer, an embedding layer, a plurality of non-linear activation layers, and a cosine similarity layer.
3. The system of claim 2 , wherein the plurality of non-linear activation layers have an activation unit of tan h, wherein tan h is a hyperbolic tangent function.
4. The system of claim 2 , wherein the hashing layer converts incoming text into multiple n-letter tokens.
5. The system of claim 3 , wherein the embedding layer represents the multiple n-letter tokens in a vector space.
6. The system of claim 1 , wherein the deep semantic similarity neural network is a Siamese network.
7. A computerized method, comprising:
obtaining training data pertaining to sample job posting search queries, the training data comprising sample job posting search results and indications as to which of the sample job posting search results were selected by members performing job posting search queries corresponding to the sample job posting search results;
for each of the sample job posting search queries, feeding the training data corresponding to the sample job posting into a machine learning algorithm to train a deep semantic similarity neural network to output a semantic similarity score indicating semantic similarity between a candidate job search query and a candidate job search result, wherein the deep semantic similarity neural network is a Siamese network;
identifying a plurality of job search results in response to a first job search query;
for each of the plurality of job search results:
feeding the job search result and the first job search query into the deep semantic similarity neural network to obtain a semantic similarity score indicating semantic similarity between the job search result and the first job search query;
ranking the plurality of job search results based on the semantic similarity score for each of the job search results;
transmitting the ranked plurality of job search results to a client device for display.
8. The method of claim 7 , wherein the deep semantic similarity neural network includes a hashing layer, an embedding layer, a plurality of non-linear activation layers, and a cosine similarity layer.
9. The method of claim 8 , wherein the plurality of non-linear activation layers have an activation unit of tan h, wherein tan h is a hyperbolic tangent function.
10. The method of claim 8 , wherein the hashing layer converts incoming text into multiple n-letter tokens.
11. The method of claim 8 , wherein the embedding layer represents the multiple n-letter tokens in a vector space.
12. The method of claim 7 , further comprising: augmenting the first job search query using terms that are semantically similar to terms in the first job search query, utilizing a semantic similarity score output from the deep semantic similarity neural network.
13. A non-transitory machine-readable storage medium comprising instructions, which when implemented by one or more machines, cause the one or more machines to perform operations comprising:
obtaining training data pertaining to sample job posting search queries, the training data comprising sample job posting search results and indications as to which of the sample job posting search results were selected by members performing job posting search queries corresponding to the sample job posting search results;
for each of the sample job posting search queries, feeding the training data corresponding to the sample job posting into a machine learning algorithm to train a deep semantic similarity neural network to output a semantic similarity score indicating semantic similarity between a candidate job search query and a candidate job search result,
identifying a plurality of job search results in response to a first job search query;
for each of the plurality of job search results:
feeding the job search result and the first job search query into the deep semantic similarity neural network to obtain a semantic similarity score indicating semantic similarity between the job search result and the first job search query;
ranking the plurality of job search results based on the semantic similarity score for each of the job search results;
transmitting the ranked plurality of job search results to a client device for display; and
augment the first job search query using terms that are semantically similar to terms in the first job search query, utilizing a semantic similarity score output from the deep semantic similarity neural network.
14. The non-transitory machine-readable storage medium of claim 13 , wherein the deep semantic similarity neural network includes a hashing layer, an embedding layer, a plurality of non-linear activation layers, and a cosine similarity layer.
15. The non-transitory machine-readable storage medium of claim 14 , wherein the plurality of non-linear activation layers have an activation unit of tan h, wherein tan h is a hyperbolic tangent function.
16. The non-transitory machine-readable storage medium of claim 14 , wherein the hashing layer converts incoming text into multiple n-letter tokens.
17. The non-transitory machine-readable storage medium of claim 14 , wherein the embedding layer represents the multiple n-letter tokens in a vector space.