1. A method of predicting the class of future customer calls to a call center, said method comprising:
providing a graphical user interface displaying prior call information comprising a call summary field including an assigned call class;
analyzing data in said call summary field using a tokenizer, said tokenizer transforming a sequence of characters in said call summary field into a sequence of tokens, producing tokenized call data;
defining a maximum entropy (MaxEnt) model producing a probability distribution of all classes for a next call to said call center based on said prior call information;
estimating parameters of said MaxEnt model for each agent of said call center based on said tokenized call data;
training a conditional random field (CRF) classifier using chronologically ordered sequences of prior calls to said call center, said CRF classifier predicting a class for a new call to said call center based on said prior call information;
receiving a new call from a returning customer;
obtaining a call class prediction for said new call being received from said returning customer based on said CRF classifier and said MaxEnt model; and
displaying said call class prediction for said new call being received from said returning customer on said graphical user interface.
2. The method according to claim 1, said prior call information further comprising contextual call information, demographics features, and time sequence of received calls.
3. The method according to claim 1, said estimating parameters of said MaxEnt model for each agent of said call center further comprising:
building a training corpus for a specific agent, said training corpus accounting for writing styles of said specific agent in order to discriminate relevant features in textual fields;
training a MaxEnt classifier to predict the class for a next call based on call summary information for a given call, the call class for said given call, and said specific agent; and
computing an Inverse Document Frequency (IDF) vector for said specific agent based on said training corpus.
4. The method according to claim 1, further comprising:
after handling said new call received from said returning customer, assigning a call class for said new call received from said returning customer.
5. The method according to claim 4, further comprising:
comparing the call class assigned for said new call received from said returning customer to said call class prediction for said call received from said returning customer;
determining prediction accuracy based on said comparing; and
adjusting said MaxEnt model and said CRF classifier when said prediction accuracy is below a predetermined threshold.
6. The method according to claim 4, further comprising:
said CRF classifier predicting a class for a next call to said call center based on said prior call information including the assigned call class for said new call received from said returning customer.
7. A method, comprising:
analyzing saved call data using a tokenizer, said tokenizer transforming a sequence of characters in a call summary field of said saved call data into a sequence of tokens, producing tokenized call data;
creating a maximum entropy (MaxEnt) model based on said tokenized call data, said MaxEnt model producing a probability distribution of all classes for a next call to a call center;
training a conditional random field (CRF) classifier with said MaxEnt model and information from said saved call data, said CRF classifier using chronologically ordered sequences of prior calls to said call center and predicting a class for a new call to said call center based on said saved call data; and
producing a call class prediction for said new call received from a returning customer based on said CRF classifier and said MaxEnt model.
8. The method according to claim 7, said information from said saved call data further comprising contextual call information, demographics features, and time sequence of received calls.
9. The method according to claim 7, said MaxEnt model being obtained for each agent of a call center.
10. The method according to claim 9, further comprising:
estimating parameters of said MaxEnt model for each agent of said call center, said estimating further comprising:
building a training corpus for a specific agent, said training corpus accounting for writing styles of said specific agent in order to discriminate relevant features in textual fields;
training a MaxEnt classifier to predict the class for a next call based on call summary information for a given call, the call class for said given call, and said specific agent; and
computing an Inverse Document Frequency (IDF) vector for said specific agent based on said training corpus.
11. The method according to claim 9, further comprising:
an agent handling said call received from said returning customer designating an assigned call class for said new call received from said returning customer.
12. The method according to claim 11, further comprising:
comparing said assigned call class for said new call received from said returning customer to said call class prediction for said call received from said returning customer;
determining prediction accuracy based on said comparing; and
adjusting said MaxEnt model and said CRF classifier when said prediction accuracy is below a predetermined threshold.
13. The method according to claim 11, further comprising:
said CRF classifier predicting a class for a next call to said call center based on said saved call data including said assigned call class for said new call received from said returning customer.
14. A system for processing communications in a call center, comprising:
a database storing records of calls to said call center, said records comprising call information comprising a call summary field that includes an assigned call class;
a special-purpose telecommunications processor connected to said database, said special-purpose telecommunications processor comprising a tokenizer; and
a graphical user interface connected to said special-purpose telecommunications processor,
said special-purpose telecommunications processor analyzing said call summary field in said records of calls using said tokenizer, said tokenizer transforming a sequence of characters in said call summary field into a sequence of tokens, producing tokenized call data,
said special-purpose telecommunications processor creating a maximum entropy (MaxEnt) model based on said tokenized call data, said MaxEnt model producing a probability distribution of all classes for a next call to said call center,
said special-purpose telecommunications processor training a conditional random field (CRF) classifier with said MaxEnt model and said call information from said records of calls, said CRF classifier using chronologically ordered sequences of prior calls to said call center and predicting a class for a new call to said call center based on said records of calls to said call center stored in said database,
said special-purpose telecommunications processor producing a call class prediction for a received call based on said CRF classifier and said MaxEnt model, and
said special-purpose telecommunications processor displaying said call class prediction for said received call on said graphical user interface.
15. The system according to claim 14, said call information further comprising:
date and time of a received call;
identity of agent handling said received call;
identity of customer making said received call;
demographics associated with said customer making said received call; and
assigned class of said received call.
16. The system according to claim 14, said special-purpose telecommunications processor creating a maximum entropy (MaxEnt) model and CRF classifier for each agent of said call center.
17. The system according to claim 16, further comprising:
said special-purpose telecommunications processor building a training corpus for a specific agent;
said special-purpose telecommunications processor training a MaxEnt classifier to predict the class for a next call based on call summary information for a given call, the call class for said given call, and said specific agent; and
said special-purpose telecommunications processor computing an Inverse Document Frequency (IDF) vector for said specific agent based on said training corpus.
18. The system according to claim 14, further comprising:
after handling said received call, said special-purpose telecommunications processor obtaining a call class assignment for said received call; and
storing the assigned class for said received call in said database.
19. The system according to claim 18, further comprising:
said special-purpose telecommunications processor comparing said call class assignment for said received call to said call class prediction for said received call;
said special-purpose telecommunications processor determining prediction accuracy based on said comparing; and
said special-purpose telecommunications processor adjusting said MaxEnt model and said CRF classifier when said prediction accuracy is below a predetermined threshold.
20. The system according to claim 18, further comprising:
said CRF classifier predicting a class for a next call to said call center based on said records of calls to said call center stored in said database including the assigned call class for said received call.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. An electronic card game playable in an amusement device having a display, the card game comprising:
(a) an electronic deck of cards having a plurality of individual playing cards, each of the individual playing cards having an associated face value;
(b) a playing area within the display;
(c) a first layer of cards of the deck being disposed on the playing area;
(d) a second layer of cards of the deck being disposed on the playing area in overlying relationship to the first layer of cards, at least one card of the second layer of cards partially covering at least three cards of the first layer of cards;
(e) a shoe of cards configured to accept cards of the deck that have been selectively removed from the playing area by a player, the shoe being disposed proximate to or within the playing area; and
(f) a set of rules dictating scoring of the electronic card game, the rules including:
(i) selecting cards from the playing area that are not covered by other cards and that have a face value that is one face value higher or one face value lower than the face value of the last card showing in the shoe;
(ii) scoring points based upon a run of play, a run of play being a duration of time when the player can selectively remove cards that are one face value higher or one face value lower than the face value of the last card in the shoe;
(iii) permitting a card in the first layer to become playable when all of the cards of the second layer which were partially covering the card in the first layer have been removed; and
(iv) allowing, in at least some instances, three or more cards in the first layer to become playable when one card of the second layer is removed.
2. The electronic card game system of claim 1, wherein the cards have conventional face values including Ace, two, three, four, five, six, seven, eight, nine, ten, Jack, Queen and King.
3. The electronic card game of claim 2, wherein the cards further include denominations including Hearts, Diamonds, Clubs and Spades.
4. The electronic card game of claim 1, further comprising:
(g) a total score indicator being disposed proximate to or within the playing area, the total score being incremented or decremented based upon the number of cards selected by the player in a run.
5. The electronic card game of claim 4, wherein the total score is also a function of the amount of time the player takes to select the next playable card.
6. The electronic card game of claim 1, further comprising:
(g) a wildcard being disposed proximate to or within the playing area and separate from the layers of cards and the shoe, the wildcard being selectable by the player when there are no playable cards that are one face value higher or one face value lower than the face value of the last card in the shoe.
7. The electronic card game of claim 1, further comprising:
(g) a wildcard being disposed proximate to or within the playing area and separate from the layers of cards and the shoe, the wildcard being selectable by the player at anytime thereby permitting the player to select any playable card.
8. The electronic card game of claim 1, further comprising:
(g) a plurality of decks of cards forming additional layers in overlying relationship relative to the second layer of cards.
9. The electronic card game of claim 1, wherein the cards have a polygonal shape with more than four sides to permit overlying relationship with more than four cards.
10. The electronic card game of claim 1, wherein the cards have a shape selected from the group consisting of a diamond shape, a square shape, a rectangular shape, a circular shape, a pentagonal shape, a octagonal shape, and a non-linear shape.
11. A method of playing an electronic card game in an amusement device having a display in accordance with a set of rules, the electronic card game including a playing area within the display and an electronic deck of cards having a plurality of individual playing cards, each of the individual playing cards having an associated face value, the method comprising:
(a) placing a first layer of cards of the deck on the playing area;
(b) placing a second layer of cards of the deck on the playing area in overlying relationship to the first layer of cards so that at least one card of the second layer of cards partially covers at least three cards of the first layer of cards;
(c) providing a shoe of cards configured to accept cards of the deck that have been selectively removed from the playing area by a player, the shoe being disposed proximate to or within the playing area; and
(d) dictating scoring of the electronic card game based upon the rules, the rules including:
(i) selecting cards from the playing area that are not covered by other cards and that have a face value that is one face value higher or one face value lower than the face value of the last card showing in the shoe;
(ii) scoring points based upon a run of play, a run of play being a duration of time when the player can selectively remove cards that are one face value higher or one face value lower than the face value of the last card in the shoe;
(iii) permitting a card in the first layer to become playable when all of the cards of the second layer which were partially covering the card in the first layer have been removed; and
(iv) allowing, in at least some instances, three or more cards in the first layer to become playable when one card of the second layer is removed.
12. The method of playing an electronic card game according to claim 11, wherein the cards have conventional face values including ace, two, three, four, five, six, seven, eight, nine, ten, jack, queen and king.
13. The method of playing an electronic card game according to claim 12, wherein the cards further include denominations including hearts, diamonds, clubs and spades.
14. The method of playing an electronic card game according to claim 11, further comprising:
(g) a total score indicator being disposed proximate to or within the playing area, the total score being incremented or decremented based upon the number of cards selected by the player in a run.
15. The method of playing an electronic card game according to claim 14, wherein the total score is also a function of the amount of time the player takes to select the next playable card.
16. The method of playing an electronic card game according to claim 11, wherein the game further comprises a wildcard being disposed proximate to or within the playing area and separate from the layers of cards and the shoe, the wildcard being selectable by the player when there are no playable cards that are one face value higher or one face value lower than the face value of the last card in the shoe.
17. The method of playing an electronic card game according to claim 11, wherein the game further comprises a wildcard being disposed proximate to or within the playing area and separate from the layers of cards and the shoe, the wildcard being selectable by the player at anytime thereby permitting the player to play any playable card.
18. The method of playing an electronic card game according to claim 11, wherein the game further comprises a plurality of decks of cards forming additional layers in overlying relationship relative to the second layer of cards.
19. The method of playing an electronic card game according to claim 11, wherein the cards have a polygonal shape with more than four sides to permit overlying relationship with more than four cards.
20. The method of playing an electronic card game according to claim 11, wherein the cards have a shape selected from the group consisting of a diamond shape, a square shape, a rectangular shape, a circular shape, a pentagonal shape, a octagonal shape, and a non-linear shape.
21. The method of playing an electronic card game according to claim 11, further comprising:
(e) placing a third layer of cards of the deck on the playing area in overlying relationship to the second layer of cards so that at least one card of the third layer of cards partially covers at least three cards of the second layer of cards.
22. An electronic game playable in an amusement device having a display, the game comprising:
(a) an electronic deck of gamepieces having a plurality of individual gamepieces, each of the individual gamepieces having an associated face value and at least three of the gamepieces having different face values;
(b) a playing area within the display;
(c) a first layer of gamepieces of the deck being disposed on the playing area;
(d) a second layer of gamepieces of the deck being disposed on the playing area in overlying relationship to the first layer of gamepieces, at least one gamepieces of the second layer of gamepieces partially covering at least three gamepieces of the first layer of gamepieces;
(e) a shoe of gamepieces configured to accept gamepieces of the deck that have been selectively removed from the playing area by a player, the shoe being disposed proximate to or within the playing area; and
(f) a set of rules dictating scoring of the electronic game, the rules including:
(i) selecting gamepieces from the playing area that are not covered by other gamepieces and that have a face value that is one face value higher or one face value lower than the face value of the last gamepieces showing in the shoe;
(ii) scoring points based upon a run of play, a run of play being a duration of time when the player can selectively remove gamepieces that are one face value higher or one face value lower than the face value of the last gamepieces in the shoe;
(iii) permitting a gamepieces in the first layer to become playable when all of the gamepieces of the second layer which were partially covering the gamepieces in the first layer have been removed; and
(iv) allowing, in at least some instances, three or more gamnepieces in the first layer to become playable when one gamepieces of the second layer is removed.
23. The electronic game of claim 22, further comprising:
(g) a total score indicator being disposed proximate to or within the playing area, the total score being incremented or decremented based upon the number of gamepieces selected by the player in a run.
24. The electronic game of claim 23, wherein the total score is also a function of the amount of time the player takes to select the next playable gamepiece.
25. The electronic game of claim 22, further comprising:
(g) a wild gamepieces being disposed proximate to or within the playing area and separate from the layers of gamepieces and the shoe, the wild gamepieces being selectable by the player when there are no playable gamepieces that are one face value higher or one face value lower than the face value of the last gamepieces in the shoe.
26. The electronic game of claim 22, further comprising:
(g) a wild gamepieces being disposed proximate to or within the playing area and separate from the layers of gamepieces and the shoe, the wild gamepieces being selectable by the player at anytime thereby permitting the player to select any playable gamepiece.
27. The electronic game of claim 22, further comprising:
(g) a plurality of decks of gamepieces forming additional layers in overlying relationship relative to the second layer of gamepieces.
28. The electronic game of claim 22, wherein the gamepieces have a polygonal shape with more than four sides to permit overlying relationship with more than four gamepieces.
29. The electronic game of claim 22, wherein the gamepieces have a shape selected from the group consisting of a diamond shape, a square shape, a rectangular shape, a circular shape, a pentagonal shape, a octagonal shape, and a non-linear shape.