Farm data annotation and bias detection

1. A method, comprising:

obtaining information related to farming activities of a farmer;

predicting an annotation category, identifying a topic, for the information by using a supervised classifier that predicts an annotation type for the information, determines, based upon a comparison against secondary sources, whether the information is important, and identifies, for information determined as important, a topic of the information using the secondary sources;

selecting an annotator for annotating the information based upon the annotation category, wherein the selecting comprises utilizing (i) a social proximity constraint identifying a social connection between the farmer and another farmer within a predetermined threshold and (ii) a farm signature constraint identifying a similarity of the farmer to another farmer exceeding a predetermined threshold, wherein the selecting an annotator comprises predicting a bias score for each of a plurality of annotators utilizing the social proximity constraint and the farm signature constraint and utilizing the bias score in the selecting an annotator;

assigning the annotator to annotate the obtained information; and

receiving annotations for the information.

2. The method of claim 1 , comprising identifying a subset of the information as relevant to a use case for the information utilizing the annotation category, wherein the identifying comprises (i) accessing at least one secondary source related to the use case and (ii) identifying relevant information based upon information identified as important within the at least one secondary source.

3. The method of claim 1 , comprising weighting the received annotation based upon (i) the social proximity constraint and (ii) the farm signature constraint.

4. The method of claim 1 , wherein the predicting comprises (i) utilizing a classifier to predict an annotation for the information, (ii) comparing the predicted annotation to a received annotation, and (iii) assigning a bias value to the received annotation.

5. The method of claim 1 , wherein the social proximity constraint is represented in a graph, the graph having a plurality of nodes and edges, each node representing a farmer, and each edge representing a social connection between two nodes connected by the edge.

6. The method of claim 1 , wherein the assigning comprises assigning a plurality of annotators responsive to determining that an annotator has historically provided biased annotations.

7. An apparatus, comprising:

at least one processor; and

a computer readable storage medium having computer readable program code embodied therewith and executable by the at least one processor, the computer readable program code comprising:

computer readable program code configured to obtain information related to farming activities of a farmer;

computer readable program code configured to predict an annotation category, identifying a topic, for the information by using a supervised classifier that predicts an annotation type for the information, determines, based upon a comparison against secondary sources, whether the information is important, and identifies, for information determined as important, a topic of the information using the secondary sources;

computer readable program code configured to select an annotator for annotating the information based upon the annotation category, wherein the selecting comprises utilizing (i) a social proximity constraint identifying a social connection between the farmer and another farmer within a predetermined threshold and (ii) a farm signature constraint identifying a similarity of the farmer to another farmer exceeding a predetermined threshold, wherein the selecting an annotator comprises predicting a bias score for each of a plurality of annotators utilizing the social proximity constraint and the farm signature constraint and utilizing the bias score in the selecting an annotator;

computer readable program code configured to assign the annotator to annotate the obtained information; and

computer readable program code configured to receive annotations for the information.

8. A computer program product, comprising:

a computer readable storage medium having computer readable program code embodied therewith, the computer readable program code executable by a processor and comprising:

computer readable program code configured to obtain information related to farming activities of a farmer;

computer readable program code configured to predict an annotation category, identifying a topic, for the information by using a supervised classifier that predicts an annotation type for the information, determines, based upon a comparison against secondary sources, whether the information is important, and identifies, for information determined as important, a topic of the information using the secondary sources;

computer readable program code configured to select an annotator for annotating the information based upon the annotation category, wherein the selecting comprises utilizing (i) a social proximity constraint identifying a social connection between the farmer and another farmer within a predetermined threshold and (ii) a farm signature constraint identifying a similarity of the farmer to another farmer exceeding a predetermined threshold, wherein the selecting an annotator comprises predicting a bias score for each of a plurality of annotators utilizing the social proximity constraint and the farm signature constraint and utilizing the bias score in the selecting an annotator;

computer readable program code configured to assign the annotator to annotate the obtained information; and

computer readable program code configured to receive annotations for the information.

9. The computer readable program code of claim 8 , comprising identifying a subset of the information as relevant to a use case for the information utilizing the annotation category, wherein the identifying comprises (i) accessing at least one secondary source related to the use case and (ii) identifying relevant information based upon information identified as important within the at least one secondary source.

10. The computer readable program code of claim 8 , comprising weighting the received annotation based upon (i) the social proximity constraint and (ii) the farm signature constraint.

11. The computer readable program code of claim 10 , wherein the predicting comprises (i) utilizing a classifier to predict an annotation for the information, (ii) comparing the predicted annotation to a received annotation, and (iii) assigning a bias value to the received annotation.

12. The computer readable program code of claim 8 , wherein the social proximity constraint is represented in a graph, the graph having a plurality of nodes and edges, each node representing a farmer, and each edge representing a social connection between two nodes connected by the edge.

13. The computer readable program code of claim 8 , wherein the assigning comprises assigning a plurality of annotators responsive to determining that an annotation has historically provided biased annotations.

14. A method comprising;

obtaining information related to farming activities of a farmer;

determining a degree of bias of each of a plurality of labelers, wherein the determining comprises utilizing a (i) social proximity constraint graph identifying social connections between the farmer and another farmer within a predetermined threshold and (ii) a farm signature constraint graph identifying a similarity of the farmer to another farmer exceeding a predetermined threshold, to predict a bias of each of the plurality of labelers;

predicting, using a supervised classifier, an annotation type for the information by determining, using the supervised classifier and based upon a comparison against secondary sources, whether the information is important, and identifying, for information determined as important, a topic of the information using the secondary sources;

selecting, based upon the degree of bias and the annotation type, a subset of the plurality of labelers to label the information; and

receiving, from the subset of the plurality of labelers, labels for the information.