Combining topic suggestions from different topic sources to assign to textual data items

• US 9,400,840 B2
• Filed: 03/21/2014
• Issued: 07/26/2016
• Est. Priority Date: 03/25/2013
• Status: Active Grant

• Current Assignee: Salesforce.com, Inc.
• Original Assignee: Salesforce.com, Inc.
• Inventors: Palmert, Joel
• Primary Examiner(s): Ho, Binh V
• Application Number:
  US14/221,971
• Publication Number:
  US 20140289231A1
• Time in Patent Office: 858 Days
• Field of Search: 707/723
• US Class Current: 1/1
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 16/345 Summarisation for human users
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

Disclosed are systems, apparatus, methods and computer-readable storage media for combining topic suggestions from different topic sources to assign to textual data items. In some implementations, a first automated topic source provides one or more first topic suggestions to associate with a textual data item. Each first topic suggestion has a respective first characteristic, which can be adjusted. A second automated topic source provides one or more second topic suggestions to associate with the textual data item. Each second topic suggestion has a respective second characteristic, which can be adjusted. The first and second topic suggestions are combined to define a combined list. A reference value is determined or retrieved to impact ordering of the topic suggestions in the combined list.

Claims

1. A computer implemented method for combining topic suggestions from different topic sources to assign to textual data items, the method comprising:

receiving, from a first automated topic source, one or more first topic suggestions to associate with a textual data item, each first topic suggestion having a respective first characteristic;

adjusting, by a processor of a computing device, the first characteristic of each first topic suggestion, comprising;

applying a first weight to each first topic suggestion;

receiving, from a second automated topic source, one or more second topic suggestions to associate with the textual data item, each second topic suggestion having a respective second characteristic;

adjusting the second characteristic of each second topic suggestion, comprising;

applying a second weight to each second topic suggestion;

combining the first and second topic suggestions to define a combined list of topic suggestions;

providing the combined list of topic suggestions suitable for a display device capable of displaying the combined list in a user interface;

receiving, at the processor from the display device, a selection of one or more of the topic suggestions from the combined list to assign to the textual data item; and

determining or retrieving a reference value to impact ordering of the topic suggestions in the combined list.

2. The method of claim 1, wherein the first characteristic is a first score, and the second characteristic is a second score.

3. The method of claim 2, wherein:

the first and second weights are stored on a storage medium in communication with the processor.

4. The method of claim 3, wherein the topic suggestions in the combined list are ordered according to weighted score.

5. The method of claim 3, wherein the one or more first topic suggestions and the one or more second topic suggestions both include a common topic suggestion, and combining the first and second topic suggestions includes summing the respective weighted scores for the common topic suggestion.

6. The method of claim 3, wherein the reference value is determined or retrieved by:

determining or retrieving an average accept ratio, the average accept ratio being a ratio of a number of selected topic suggestions from the combined list to a number of total topic suggestions in the combined list;

determining or retrieving an accept ratio for each topic source, the accept ratio being a ratio of a number of selected topic suggestions from the topic source to a number of topic suggestions from the topic source; and

adjusting one or more of the weights stored on the storage medium according to a comparison of the accept ratio for a corresponding topic source with the average accept ratio.

7. The method of claim 6, wherein adjusting a weight includes increasing or decreasing a previous value of the weight proportionally with respect to a control ratio of the accept ratio for the corresponding topic source to the average accept ratio.

8. The method of claim 7, wherein the weight is increased when the control ratio is greater than one, and the weight is decreased when the control ratio is less than one.

9. The method of claim 7, wherein the weight is adjusted from the previous value to a current value by applying the control ratio to the previous value.

10. The method of claim 7, wherein an exponential factor is applied to the control ratio, the exponential factor governing a rate of the adjustment of the weight.

11. The method of claim 6, wherein the average accept ratio and the accept ratio for each topic source are determined, and the one or more weights are adjusted, over a designated interval.

12. The method of claim 6, wherein adjusting a weight includes increasing or decreasing a previous value of the weight proportionally with respect to the number of selected topic suggestions from the corresponding topic source.

13. The method of claim 6, wherein adjusting a weight includes resetting the weight to a default value.

14. The method of claim 1, further comprising:

determining that the number of selected topic suggestions from the topic source meets or exceeds a threshold.

15. The method of claim 1, wherein:

each topic source is one of;

a topic index, a natural language processing (NLP) library, a consecutive words identifier, a noun phrase extractor, and a use-based word identifier, and

the textual data item includes textual content and is one of;

a social networking message, a database record, a feed-based record update, a document, a knowledge article, a scientific paper, a book, a web page, an online encyclopedia entry, a database record, a customer relationship management (CRM) object, a user profile, a group file, an email message, an instant message, and a text message.

16. The method of claim 1, further comprising:

associating the selected one or more topic suggestions with the textual data item, including updating a field in a database system storing the textual data item to indicate the association.

17. The method of claim 16, wherein the textual data item is an information update, and the selected one or more topic suggestions are capable of being displayed in the user interface in conjunction with a social network feed including the information update.

18. The method of claim 1, wherein the display device is one of:

a smartphone, a laptop, a tablet, a wearable display device, a desktop computer, and a monitor.

19. One or more computing devices for combining topic suggestions from different topic sources to assign to textual data items, the one or more computing devices comprising:

one or more processors operable to execute one or more instructions to;

receive, from a first automated topic source, one or more first topic suggestions to associate with a textual data item, each first topic suggestion having a respective first characteristic,

adjust the first characteristic of each first topic suggestion, comprising;

applying a first weight to each first topic suggestion,

receive, from a second automated topic source, one or more second topic suggestions to associate with the textual data item, each second topic suggestion having a respective second characteristic,

adjust the second characteristic of each second topic suggestion, comprising;

applying a second weight to each second topic suggestion,

combine the first and second topic suggestions to define a combined list of topic suggestions, and

determine or retrieve a reference value to impact ordering of the topic suggestions in the combined list.

20. A non-transitory computer-readable storage medium storing instructions executable by a processor to cause a method to be performed for combining topic suggestions from different topic sources to assign to textual data items, the method comprising:

receiving, from a first automated topic source, one or more first topic suggestions to associate with a textual data item, each first topic suggestion having a respective first characteristic;

adjusting the first characteristic of each first topic suggestion, comprising;

applying a first weight to each first topic suggestion;

receiving, from a second automated topic source, one or more second topic suggestions to associate with the textual data item, each second topic suggestion having a respective second characteristic;

adjusting the second characteristic of each second topic suggestion, comprising;

applying a second weight to each second topic suggestion;

combining the first and second topic suggestions to define a combined list of topic suggestions; and

determining or retrieving a reference value to impact ordering of the topic suggestions in the combined list.