Methods and systems for searching for color themes, suggesting color theme tags, and estimating tag descriptiveness
View Patent ↗Embodiments of the present disclosure assess how well a tag describes a color theme by estimating a descriptiveness value for the tag for the color theme. Some embodiments determine descriptiveness values for a tag based on weighted color attributes determined from the tag's existing use in a color theme collection. Descriptiveness values are used generally in color theme searching and to suggest tags for a color theme, among other things.
1. A method, comprising:
identifying, at a computer, at least one color attribute for a tag based on at least one association of the tag with a first color theme in a collection of color themes, the at least one color attribute being defined as present or absent for the first color theme;
determining, via the computer, a weight for the at least one color attribute based on the at least one association of the tag with the first color theme, the weight representing a significance of the at least one color attribute to the tag; and
estimating, via the computer, a descriptiveness value of the tag for a second color theme in the collection of color themes using the at least one color attribute and the weight.
2. The method of claim 1 further comprising determining strong classifier parameters, the strong classifier parameters comprising the at least one color attribute used as a weak classifier and the weight.
3. The method of claim 2 , wherein estimating the descriptiveness value comprises determining at least one weak classifier result and a strong classifier result that uses the at least one weak classifier result according to the weight.
4. The method of claim 2 , wherein the strong classifier parameters are determined by identifying one or more color themes in the collection of color themes having the tag.
5. The method of claim 1 , wherein identifying the at least one color attribute comprises using an Adaboost algorithm.
6. The method of claim 5 , wherein determining the weight comprises using the Adaboost algorithm.
7. The method of claim 1 , wherein the descriptiveness value of the tag comprises a probabilistic classifier value.
8. The method of claim 1 , wherein estimating the descriptiveness value of the tag comprises applying a Platt scaling algorithm to provide a probabilistic classifier value between 0 and 1.
9. The method of claim 1 , further comprising estimating a color-theme descriptiveness value for each of the color themes in the collection.
10. The method of claim 9 further comprising:
receiving the tag as a search term entered in a tag-based color theme search; and
suggesting one or more color themes based at least in part on the color-theme descriptiveness value of the tag for each of the color themes in the collection.
11. The method of claim 1 , wherein identifying the at least one color attribute comprises identifying at least two color attributes based on at least some color themes in the collection associated with the tag and at least some color themes in the collection not associated with the tag.
12. The method of claim 11 , wherein determining the weight for the at least one color attribute comprises determining a weight for each of the at least two color attributes based on the at least some color themes in the collection associated with the tag and the at least some color themes in the collection not associated with the tag.
13. The method of claim 12 , wherein estimating a descriptiveness value further comprises using the at least two color attributes and the weight for each of the at least two color attributes.
14. The method of claim 1 wherein the weight for each at least one color attribute is determined using a data set comprising the at least one association of the tag with the first color theme and associations of the tag with other color themes in the collection of color themes.
15. The method of claim 1 wherein the weight for each at least one color attribute is determined using a data set comprising:
positive examples of color themes in the collection of color themes associated with the tag; and
negative examples of color themes in the collection of color themes not associated with the tag.
16. The method of claim 1 wherein the color theme comprises two or more colors.
17. The method of claim 1 wherein identifying the at least one color attribute comprises using an algorithm that uses a data set comprising:
positive examples of color themes in the collection of color themes associated with the tag; and
negative examples of color themes in the collection of color themes not associated with the tag.
18. A method, comprising:
for a tag associated with a color theme in a collection of color themes:
identifying at least one color attribute based on at least one association of the tag with a first color theme in the collection of color themes, the at least one color attribute being defined as present or absent for the first color theme; and
determining a weight for each of the at least one color attributes based on the at least one association of the tag with the color theme, the weight representing a significance of the at least one color attribute to the tag;
estimating, via a computer, a descriptiveness value of the tag for an identified color theme using the at least one color attribute and the weight; and
suggesting, via the computer, the tag as an appropriate descriptor for the identified color theme if the descriptiveness value of the tag for the identified color theme is greater than a threshold value.
19. The method of claim 18 , wherein identifying the at least one color attribute and determining the weight comprises using an Adaboost algorithm.
20. The method of claim 18 , wherein estimating the descriptiveness value of each tag comprises applying a Platt scaling algorithm to provide a probabilistic classifier value between 0 and 1.
21. A method comprising:
for a search term:
identifying at least one color attribute based on at least one association of the search term with a first color theme in a collection of color themes, the at least one color attributes being defined as present or absent for the first color theme; and
determining a weight for the at least one color attribute based on the at least one association of the search term with the first color theme, the weight representing a significance of the tag for the first color theme;
estimating, via a computer, a descriptiveness value of the search term for each color theme in the collection of color themes using the at least one color attribute and the weight; and
suggesting, via the computer, at least one color theme from the collection of color themes based at least in part on the descriptiveness value of the search term for each color theme in the collection of color themes.
22. The method of claim 21 , wherein identifying the at least one color attribute and determining the weight comprises using an Adaboost algorithm.
23. The method of claim 21 , wherein estimating the descriptiveness value of the search term comprises applying a Platt scaling algorithm to provide a probabilistic classifier value between 0 and 1.
24. The method of claim 21 , wherein suggesting at least one color theme comprises suggesting up to a predetermined maximum number of color themes.
25. The method of claim 21 wherein the descriptiveness value of the search term for each color theme comprises a quantified value and wherein suggesting at least one color theme comprises selecting color themes for which the quantified value is greater than a threshold value.
26. The method of claim 21 wherein suggesting one or more color themes is based at least in part on a descriptiveness value of a second search term for each color theme.
27. A system, comprising:
a database system comprising a collection of color themes;
a computer system configured to:
identify at least one color attribute for a tag based on at least one association of the tag with a first color theme in a collection of color themes, the at least one color attribute being defined as present or absent for the first color theme;
determine a weight for the at least one color attribute based on the at least one association of the tag with the first color theme, the weight representing a significance of the at least one color attribute to the tag; and
estimate a descriptiveness value of the tag for a second color theme in the collection of color themes using the at least one color attribute and the weight; and
wherein the database system further comprises storing the color attributes, weights, and descriptiveness values.
28. A non-transitory computer-readable medium on which is encoded program code, the program code comprising:
program code for identifying at least one color attribute for a tag based on at least one association of the tag with a first color theme in a collection of color themes, the at least one color attribute being defined as present or absent for the first color theme, and determining a weight for the at least one color attribute based on the at least one association of the tag with the first color theme, the weight representing a significance of the at least one color attribute to the tag; and
program code for estimating a descriptiveness value of the tag for a second color theme in the collection of color themes using the at least one color attribute and the weight.