Evaluation score determination machine learning models with differential periodic tiers
Various embodiments of the present invention address technical challenges associated with performing machine learning operations on timeseries/periodic data by introducing a machine learning framework that has a first periodic tier for determining predicted evaluation scores for those predictive entities that are associated with a single evaluation period (e.g., a single year of data) and a second periodic tier for determining predicted evaluation scores for those predictive entities that are associated with multiple evaluation periods. The noted framework addresses the existing shortcomings of machine learning frameworks that operate on timeseries/periodic data with respect to inadequacy of data associated with shorter periods to determine parameters needed to perform comprehensive predictive data analysis with respect to longer periods.
1 . A computer-implemented method comprising:
receiving, by one or more processors, aggregated predictive entity data for a predictive entity, wherein the aggregated predictive entity data corresponds to an evaluation period count that describes a number of one or more evaluation periods for which the aggregated predictive entity data is available with respect to the predictive entity and the aggregated predictive entity data is scraped from a first engagement platform and a second engagement platform;
generating, by the one or more processors, a set of engagement attributes for the predictive entity based at least in part on the aggregated predictive entity data associated with the predictive entity, wherein:
(i) the set of engagement attributes comprise a cross-platform evaluation score, a cross-platform deviation penalty, and the evaluation period count,
(ii) the cross-platform evaluation score is generated by:
(a) determining, based at least in part on the evaluation period count, a first per-platform evaluation score for the predictive entity with respect to an evaluation period of the one or more evaluation periods and the first engagement platform,
(b) determining, based at least on part on the evaluation period count, a second per-platform evaluation score for the predictive entity with respect to the evaluation period and the second engagement platform,
(c) determining a cross-platform evaluation deviation measure for the predictive entity based at least in part on the first per-platform evaluation score and the second per-platform evaluation score, and
(d) generating the cross-platform evaluation score based at least in part on the cross-platform evaluation deviation measure; and
(iii) the cross-platform deviation penalty is generated based at least in part on whether the cross-platform evaluation deviation measure and a cross-platform deviation threshold;
generating based at least in part on the set of engagement attributes, a predicted evaluation score, wherein the predicted evaluation score comprises a selection of one of the first per-platform evaluation score or the second per-platform evaluation score in response to the cross-platform evaluation deviation measure meeting or exceeding a deviation threshold; and
updating, by the one or more processors and based at least in part on the predicted evaluation score, a position of the predictive entity within a set of search results rendered within a user interface.
2 . The computer-implemented method of claim 1 , further comprising:
determining one or more periodic per-platform evaluation scores for the first per-platform evaluation score with respect to the one or more evaluation periods; and
determining the first per-platform evaluation score based at least in part on the one or more periodic per-platform evaluation scores.
3 . The computer-implemented method of claim 1 , wherein:
the set of engagement attributes comprises an unstable engagement penalty that is determined based at least in part on whether a stable engagement score for the aggregated predictive entity data fails to satisfy a stable engagement score threshold; and
the stable engagement score is determined based at least in part on an engagement designation count of engagement designations associated with the aggregated predictive entity data.
4 . The computer-implemented method of claim 1 , wherein:
the set of engagement attributes comprises a stable sentiment bonus that is determined based at least in part on whether a stable sentiment score for the aggregated predictive entity data fails to satisfy a stable sentiment score threshold; and
the stable sentiment score is determined based at least in part on one or more sentiment values associated with the aggregated predictive entity data across one or more evaluation periods.
5 . The computer-implemented method of claim 1 , wherein:
the set of engagement attributes comprises an entity profile weightage factor that describes an inferred degree of correspondence between one or more entity attributes for the predictive entity and a preferred entity profile.
6 . The computer-implemented method of claim 1 , further comprising:
in response to determining that the predictive entity is associated with a single evaluation period, determining the predicted evaluation score based at least in part on a current period high-ranked feedback count for the predictive entity and a preceding period high-ranked feedback count for the predictive entity.
7 . A system comprising:
one or more processors; and
one or more memories storing processor-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
receiving aggregated predictive entity data for a predictive entity, wherein the aggregated predictive entity data corresponds to an evaluation period count that describes a number of one or more evaluation periods for which the aggregated predictive entity data is available with respect to the predictive entity and the aggregated predictive entity data is scraped from a first engagement platform and a second engagement platform;
generating a set of engagement attributes for the predictive entity and based at least in part on the aggregated predictive entity data associated with the predictive entity, wherein:
(i) the set of engagement attributes comprises a cross-platform evaluation score, a cross-platform deviation penalty, and the evaluation period count,
(ii) the cross-platform evaluation score is generated by:
(a) determining, based at least in part on the evaluation period count, a first per-platform evaluation score for the predictive entity with respect to an evaluation period of the one or more evaluation periods and the first engagement platform,
(b) determining, based at least in part on the evaluation period count, a second per-platform evaluation score for the predictive entity with respect to the evaluation period and the second engagement platform,
(c) determining a cross-platform evaluation deviation measure for the predictive entity based at least in part on the first per-platform evaluation score and the second per-platform evaluation score, and
(d) generating the cross-platform evaluation score based at least in part on the cross-platform evaluation deviation measure; and
(iii) the cross-platform deviation penalty is generated based at least in part on the cross-platform evaluation deviation measure and a cross-platform deviation threshold;
generating, based at least in part on the set of evaluation attributes, a predicted evaluation score, wherein the predicted evaluation score comprises a selection of one of the first per-platform evaluation score or the second per-platform evaluation score in response to the cross-platform evaluation deviation measure meeting or exceeding a deviation threshold; and
updating based at least in part on the predicted evaluation score, a position of the predictive entity within a set of search results rendered within a user interface.
8 . The system of claim 7 , wherein the operations further comprise:
determining one or more periodic per-platform evaluation scores for the first per-platform evaluation score with respect to the one or more evaluation periods; and
determining the first per-platform evaluation score based at least in part on the one or more periodic per-platform evaluation scores.
9 . The system of claim 7 , wherein:
the set of engagement attributes comprises an unstable engagement penalty that is determined based at least in part on whether a stable engagement score for the aggregated predictive entity data fails to satisfy a stable engagement score threshold; and
the stable engagement score is determined based at least in part on an engagement designation count of engagement designations associated with the aggregated predictive entity data.
10 . The system of claim 7 , wherein:
the set of engagement attributes comprises a stable sentiment bonus that is determined based at least in part on whether a stable sentiment score for the aggregated predictive entity data fails to satisfy a stable sentiment score threshold; and
the stable sentiment score is determined based at least in part on one or more sentiment values associated with the aggregated predictive entity data across one or more evaluation periods.
11 . The system of claim 7 , wherein:
the set of engagement attributes comprises an entity profile weightage factor that describes an inferred degree of correspondence between one or more entity attributes for the predictive entity and a preferred entity profile.
12 . The system of claim 7 , further comprising:
in response to determining that the predictive entity is associated with a single evaluation period, determining the predicted evaluation score based at least in part on a current period high-ranked feedback count for the predictive entity and a preceding period high-ranked feedback count for the predictive entity.
13 . One or more non-transitory computer-readable media storing processor-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
receiving aggregated predictive entity data for a predictive entity, wherein the aggregated predictive entity data corresponds to an evaluation period count that describes a number of one or more evaluation periods for which the aggregated predictive entity data is available with respect to the predictive entity and the aggregated predictive entity data is scraped from a first engagement platform and a second engagement platform;
generating a set of engagement attributes for the predictive entity and based at least in part on the aggregated predictive entity data associated with the predictive entity, wherein:
(i) the set of engagement attributes comprise a cross-platform evaluation score, a cross-platform deviation penalty, and the evaluation period count,
(ii) the cross-platform evaluation score is generated by:
(a) determining, based at least in part on the evaluation period count, a first per-platform evaluation score for the predictive entity with respect to an evaluation period of the one or more evaluation periods and the first engagement platform,
(b) determining, based at least in part on the evaluation period count, a second per-platform evaluation score for the predictive entity with respect to the evaluation period and the second engagement platform,
(c) determining a cross-platform evaluation deviation measure for the predictive entity based at least in part on the first per-platform evaluation score and the second per-platform evaluation score, and
(d) generating the cross-platform evaluation score based at least in part on the cross-platform evaluation deviation measure; and
(iii) the cross-platform deviation penalty is generated based at least in part on the cross-platform evaluation deviation measure and a cross-platform deviation threshold;
generating, based at least in part on the set of evaluation attributes, a predicted evaluation score, wherein the predicted evaluation score comprises a selection of one of the first per-platform evaluation score or the second per-platform evaluation score in response to the cross-platform evaluation deviation measure meeting or exceeding a deviation threshold; and
updating based at least in part on the predicted evaluation score, a position of the predictive entity within a set of search results rendered within a user interface.
14 . The one or more non-transitory computer-readable media of claim 13 , wherein the operations further comprise:
determining one or more periodic per-platform evaluation scores for the first per-platform evaluation score with respect to the one or more evaluation periods; and
determining the first per-platform evaluation score based at least in part on the one or more periodic per-platform evaluation scores.
15 . The one or more non-transitory computer-readable media of claim 13 , wherein:
the set of engagement attributes comprises an unstable engagement penalty that is determined based at least in part on whether a stable engagement score for the aggregated predictive entity data fails to satisfy a stable engagement score threshold; and
the stable engagement score is determined based at least in part on an engagement designation count of engagement designations associated with the aggregated predictive entity data.
16 . The one or more non-transitory computer-readable media of claim 13 , wherein:
the set of engagement attributes comprises a stable sentiment bonus that is determined based at least in part on whether a stable sentiment score for the aggregated predictive entity data fails to satisfy a stable sentiment score threshold; and
the stable sentiment score is determined based at least in part on one or more sentiment values associated with the aggregated predictive entity data across one or more evaluation periods.
17 . The one or more non-transitory computer-readable media of claim 13 , wherein:
the set of engagement attributes comprises an entity profile weightage factor that describes an inferred degree of correspondence between one or more entity attributes for the predictive entity and a preferred entity profile.