Collaborative creation of annotation training data

1. A method for collaborative annotation of a document, the method comprising:

generating, by a host computer, a graphical user interface (GUI) that displays a document to annotate;

transmitting, by the host computer to a plurality of client computers, the GUI;

establishing, by the host computer, a plurality of real time communication modes between the client computers, each communication mode including communication channels between all client computers;

receiving, by the host computer from of the plurality of client computers, a plurality of correct annotations and incorrect annotations entered into the GUIs at the client computers, each correct annotation and incorrect annotation identifying a span of words in the document to annotate and including an associated annotation type, the correct annotations and incorrect annotations identified by a coordinator;

identifying, by the host computer, conflicts between the received correct annotations and incorrect annotations, wherein a conflict is an associated type of one of the annotations does not match the associated type of another annotation or the span of words is associated with fewer annotations than the number of the plurality of the client computers;

generating, by the host computer, a summary by type including each correct and incorrect annotation;

updating, by the host computer, in real time, the GUI to include for display the received plurality of correct and incorrect annotations and inter-rater agreement statistics indicating the level of agreement between the correct and incorrect annotations received from each client computer and further including for display the identified conflicts between the received correct and incorrect annotations; and

transmitting, by the host computer to the plurality of client computers, the updated GUI.

2. The method in accordance with claim 1 , further comprising:

receiving, by the host computer from one of the plurality of client computers, an indication that all correct annotations have been received; and

writing to a datastore in a desired document format or formats, a correctly annotated document, including at least all the correct annotations to the span of words.

3. The method in accordance with claim 2 , further comprising:

storing, by the host computer, all communications occurring over the plurality of real time communication modes between the client computers;

in response to receiving, by the host computer, a request to generate a timing mark, storing an association of an annotation and a timestamp of the real time communications;

writing to a datastore a guideline outline that includes each correct annotation and incorrect annotation to the span of words or links to any associated real time discussions that may have occurred over the multimodal communication interfaces; and

wherein writing to a datastore the annotated document further includes writing the associations of the annotations and timestamps of the real time communications.

4. The method in accordance with claim 1 , wherein the GUI is a web page that includes embedded scripts to receive annotations of the document to annotate from a user of a client computer and transmit the annotations to the host computer.

5. A computer program product for collaborative annotation of a document, the computer program product comprising:

one or more non-transitory computer-readable storage media and program instructions stored on the one or more non-transitory computer-readable media, the program instructions comprising:

program instructions to generate, by a host computer, a graphical user interface (GUI) that displays a document to annotate;

program instructions to transmit, by the host computer to a plurality of client computers, the GUI;

program instructions to establish, by the host computer, a plurality of real time communication modes between the client computers, each communication mode including communication channels between all client computers;

program instructions to receive, by the host computer from of the plurality of client computers, a plurality of correct annotations and incorrect annotations entered into the GUIs at the client computers, each correct annotation and incorrect annotation identifying a span of words in the document to annotate and including an associated annotation type, the correct annotations and incorrect annotations identified by a coordinator;

program instructions to identify, by the host computer, conflicts between the received correct annotations and incorrect annotations, wherein a conflict is an associated type of one of the annotations does not match the associated type of another annotation or the span of words is associated with fewer annotations than the number of the plurality of the client computers;

program instructions to generate, by the host computer, a summary by type including each correct and incorrect annotation;

program instructions to update, by the host computer, in real time, the GUI to include for display the received plurality of correct and incorrect annotations and inter-rater agreement statistics indicating the level of agreement between the correct and incorrect annotations received from each client computer and further including for display the identified conflicts between the received correct and incorrect annotations; and

program instructions to transmit, by the host computer to the plurality of client computers, the updated GUI.

6. The computer program product in accordance with claim 5 , further comprising:

program instructions to receive, by the host computer from one of the plurality of client computers, an indication that all correct annotations have been received; and

program instructions to write to a datastore in a desired document format or formats, a correctly annotated document, including at least all the correct annotations to the span of words.

7. The computer program product in accordance with claim 6 , further comprising:

program instructions to store, by the host computer, all communications occurring over the plurality of real time communication modes between the client computers;

program instructions, in response to receiving, by the host computer, a request to generate a timing mark, to store an association of an annotation and a timestamp of the real time communications;

program instructions to write to a datastore a guideline outline that includes each correct annotation and incorrect annotation to the span of words or links to any associated real time discussions that may have occurred over the multimodal communication interfaces; and

wherein the program instructions to write to the datastore the annotated document further includes program instructions to write the associations of the annotations and timestamps of the real time communications.

8. The computer program product in accordance with claim 5 , wherein the GUI is a web page that includes embedded scripts to receive annotations of the document to annotate from a user of a client computer and transmit the annotations to the host computer.

9. A computer system for collaborative annotation of a document, the computer system comprising:

one or more computer processors, one or more computer-readable storage media, and program instructions stored on one or more off the computer-readable media for execution by at least one of the one or more processors, the program instructions comprising:

program instructions to generate, by a host computer, a graphical user interface (GUI) that displays a document to annotate;

program instructions to transmit, by the host computer to a plurality of client computers, the GUI;

program instructions to establish, by the host computer, a plurality of real time communication modes between the client computers, each communication mode including communication channels between all client computers;

program instructions to receive, by the host computer from of the plurality of client computers, a plurality of correct annotations and incorrect annotations entered into the GUIs at the client computers, each correct annotation and incorrect annotation identifying a span of words in the document to annotate and including an associated annotation type, the correct annotations and incorrect annotations identified by a coordinator;

program instructions to identify, by the host computer, conflicts between the received correct annotations and incorrect annotations, wherein a conflict is an associated type of one of the annotations does not match the associated type of another annotation or the span of words is associated with fewer annotations than the number of the plurality of the client computers;

program instructions to generate, by the host computer, a summary by type including each correct and incorrect annotation;

program instructions to update, by the host computer, in real time, the GUI to include for display the received plurality of correct and incorrect annotations and inter-rater agreement statistics indicating the level of agreement between the correct and incorrect annotations received from each client computer and further including for display the identified conflicts between the received correct and incorrect annotations; and

program instructions to transmit, by the host computer to the plurality of client computers, the updated GUI.

10. The computer system in accordance with claim 9 , further comprising:

program instructions to receive, by the host computer from one of the plurality of client computers, an indication that all correct annotations have been received; and

program instructions to write to a datastore in a desired document format or formats, a correctly annotated document, including at least all the correct annotations to the span of words.

11. The computer system in accordance with claim 10 , further comprising:

program instructions to store, by the host computer, all communications occurring over the plurality of real time communication modes between the client computers;

program instructions, in response to receiving, by the host computer, a request to generate a timing mark, to store an association of an annotation and a timestamp of the real time communications;

program instructions to write to a datastore a guideline outline that includes each correct annotation and incorrect annotation to the span of words or links to any associated real time discussions that may have occurred over the multimodal communication interfaces; and

wherein the program instructions to write to the datastore the annotated document further includes program instructions to write the associations of the annotations and timestamps of the real time communications.

12. The computer system in accordance with claim 9 , wherein the GUI is a web page that includes embedded scripts to receive annotations of the document to annotate from a user of a client computer and transmit the annotations to the host computer.