Methods, apparatuses and computer program products for improving data retrieval relevance and reducing data retrieval latency in cloud-based platforms
Various examples herein described are related to methods, apparatuses, and computer program products for improving data retrieval relevance and reducing data retrieval latency in cloud-based issue and page platforms are provided. For example, an example computer-implemented method may comprise receiving query content metadata through a query input user interface; generating one or more query variant data objects; retrieving one or more relevant page data objects from a plurality of page data objects; generating one or more answer data objects; and rendering the one or more answer data objects on an answer output user interface.
1 . An apparatus comprising at least one processor and at least one non-transitory memory comprising program code, the at least one non-transitory memory and the program code configured to, with the at least one processor, cause the apparatus to at least:
receive query content metadata via a user interface of a cloud-based issue and page platform;
retrieve, based on user identifier metadata associated with the query content metadata, a user data object corresponding to a user;
determine user authority metadata from the user data object, the user authority metadata indicating whether the user is an administrator;
generate one or more query variant data objects using a query variant formulation layer;
select one or more data object repositories based on the user authority metadata, wherein:
when the user authority metadata indicates that the user is the administrator, the one or more selected data object repositories include one or more external data object repositories, and
when the user authority metadata indicates that the user is not the administrator, the one or more selected data object repositories are restricted to data object repositories within the cloud-based issue and page platform;
transmit one or more retrieval queries corresponding to the one or more query variant data objects to only the one or more selected data object repositories;
receive identifying metadata for one or more page data objects from the one or more selected data object repositories;
retrieve one or more page data objects based on the identifying metadata;
generate one or more answer data objects based on the one or more retrieved page data objects using a query answer formulation layer; and
render the one or more answer data objects via the user interface.
2 . The apparatus of claim 1 , wherein the query variant formulation layer and the query answer formulation layer comprise one or more generative pre-trained transformers.
3 . The apparatus of claim 1 , wherein the one or more query variant data objects comprise at least one keyword-based query variant data object, wherein the at least one keyword-based query variant data object comprises extracted keyword metadata based on the query content metadata.
4 . The apparatus of claim 1 , wherein the one or more query variant data objects comprise at least one synonym-based query variant data object, wherein the at least one synonym-based query variant data object comprises predicted synonym metadata based on the query content metadata.
5 . The apparatus of claim 1 , wherein the one or more query variant data objects comprise at least one location-based query variant data object, wherein the at least one location-based query variant data object comprises predicted location metadata based on the user data object associated with the query content metadata.
6 . The apparatus of claim 1 , wherein the at least one non-transitory memory and the program code are configured to, with the at least one processor, cause the apparatus to:
transmit the one or more query variant data objects in parallel to a page data object retrieval application programming interface (API); and
receive the identifying metadata associated with the one or more page data objects from the page data object retrieval API.
7 . The apparatus of claim 1 , wherein the at least one non-transitory memory and the program code are configured to, with the at least one processor, cause the apparatus to:
generate relevant page data object ranking metadata by inputting the one or more retrieved page data objects to a plurality of encoders of a query result formulation layer; and
generate the one or more answer data objects by inputting the relevant page data object ranking metadata and the one or more retrieved page data objects to the query answer formulation layer.
8 . A computer-implemented method comprising:
receiving query content metadata via a user interface of a cloud-based issue and page platform;
retrieving, based on user identifier metadata associated with the query content metadata, a user data object corresponding to a user;
determining user authority metadata from the user data object, the user authority metadata indicating whether the user is an administrator;
generating one or more query variant data objects using a query variant formulation layer;
selecting one or more data object repositories based on the user authority metadata, wherein, in response to determining that the user authority metadata indicates that the user is not the administrator, the one or more selected data object repositories are restricted to data object repositories within the cloud-based issue and page platform;
transmitting one or more retrieval queries corresponding to the one or more query variant data objects to only the one or more selected data object repositories;
receiving identifying metadata for one or more page data objects from the one or more selected data object repositories;
retrieving one or more page data objects based on the identifying metadata;
generating one or more answer data objects based on the one or more retrieved page data objects using a query answer formulation layer; and
rendering the one or more answer data objects via the user interface.
9 . The computer-implemented method of claim 8 , wherein the query variant formulation layer and the query answer formulation layer comprise one or more generative pre-trained transformers.
10 . The computer-implemented method of claim 8 , wherein the one or more query variant data objects comprise at least one keyword-based query variant data object, wherein the at least one keyword-based query variant data object comprises extracted keyword metadata based on the query content metadata.
11 . The computer-implemented method of claim 8 , wherein the one or more query variant data objects comprise at least one synonym-based query variant data object, wherein the at least one synonym-based query variant data object comprises predicted synonym metadata based on the query content metadata.
12 . The computer-implemented method of claim 8 , wherein the one or more query variant data objects comprise at least one location-based query variant data object, wherein the at least one location-based query variant data object comprises predicted location metadata based on the user data object associated with the query content metadata.
13 . The computer-implemented method of claim 8 , wherein the computer-implemented method comprises:
transmitting the one or more query variant data objects in parallel to a page data object retrieval application programming interface (API) of the cloud-based issue and page platform; and
receiving the identifying metadata associated with the one or more retrieved page data objects from the page data object retrieval API.
14 . The computer-implemented method of claim 8 , further comprising:
generating relevant page data object ranking metadata by inputting the one or more retrieved page data objects to a plurality of encoders of a query result formulation layer; and
generating the one or more answer data objects by inputting the relevant page data object ranking metadata and the one or more retrieved page data objects to the query answer formulation layer.
15 . A computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising an executable portion configured to:
receive query content metadata via a user interface of a cloud-based issue and page platform;
retrieve, based on user identifier metadata associated with the query content metadata, a user data object corresponding to a user;
determine user authority metadata from the user data object, the user authority metadata indicating whether the user is an administrator;
generate one or more query variant data objects using a query variant formulation layer;
select one or more data object repositories based on the user authority metadata, wherein:
when the user authority metadata indicates that the user is the administrator, the one or more selected data object repositories include one or more external data object repositories, and
when the user authority metadata indicates that the user is not the administrator, the one or more selected data object repositories are restricted to data object repositories within the cloud-based issue and page platform;
transmit one or more retrieval queries corresponding to the one or more query variant data objects to only the one or more selected data object repositories;
receive identifying metadata for one or more page data objects from the one or more selected data object repositories;
retrieve one or more page data objects based on the identifying metadata;
generate one or more answer data objects based on the one or more retrieved page data objects using a query answer formulation layer; and
render the one or more answer data objects via the user interface.
16 . The computer program product of claim 15 , wherein the query variant formulation layer and the query answer formulation layer comprise one or more generative pre-trained transformers.
17 . The computer program product of claim 15 , wherein the one or more query variant data objects comprise at least one keyword-based query variant data object, wherein the at least one keyword-based query variant data object comprises extracted keyword metadata based on the query content metadata.