Method for adaptive conversation state management with filtering operators applied dynamically as part of a conversational interface
A system and method of processing a search request is provided. Identification of a desired content item is based on comparing a topic of the search request to previous user input. The method includes providing access to a set of content items with metadata that describes the corresponding content items and providing information about previous searches. The method further includes receiving a present input from the user and determining a relatedness measure between the information about the previous searches and an element of the present input. If the relatedness measure is high, the method also includes selecting a subset of content items based on comparing the present input and information about the previous searches with the metadata that describes the subset of content items. Otherwise, the method includes selecting a subset of content items based on comparing the present input with the metadata that describes the subset of content items.
1. A computer-implemented method including a hardware processor for processing search requests, the computer-implemented method comprising:
receiving, in a conversation state space defined by a first filter, the first filter comprising a first term of a first attribute type, a first speech input;
identifying in the first speech input a second filter comprising a second term of a second attribute type;
generating a relationship graph that identifies relationships between a plurality of terms, including the first and second term, wherein the relationship graph is generated by the hardware processor retrieving and processing information from repositories to identify semantic associations between the plurality of terms;
selecting, based on the relationship graph, a first set of search results, wherein each search result of the first set of search results corresponds to a respective node of the relationship graph that is connected, by an edge having an edge type corresponding to the second attribute type, to a first node corresponding to the first term of the first attribute type and having a node type corresponding to the first attribute type;
generating for display the first set of search results;
receiving, in the conversation state space, a second speech input;
identifying in the second speech input a third filter comprising a third term of a third attribute type;
determining, based on the relationship graph, whether any of the first search results match the first filter, the second filter, and the third filter;
in response to determining that no search results match: (a) the first filter and the second filter for which the first set of search results is being generated for display, and (b) the third filter which was received in the conversation state space defined by first filter and the second filter:
automatically removing the second filter from the conversation state space and replacing it in the conversation state space with the third filter;
selecting a second set of search results based on the first filter and the third filter and not the second filter; and
generating for display the second set of search results.
2. The computer-implemented method of claim 1 , wherein determining, based on the relationship graph, whether any of the first search results match the first filter, the second filter, and the third filter comprises:
determining, for each search result of the first set of search results, whether a first node of the relationship graph corresponding to a respective search result is connected to a second node of the relationship graph having a node type corresponding to the third term by an edge having an edge type corresponding to the third attribute type.
3. The computer-implemented method of claim 1 , wherein selecting the second set of search results based on the first filter and the third filter comprises:
selecting, based on the relationship graph, the second set of search results, wherein each search result of the second set of search results corresponds to a respective node of the relationship graph that is connected, by an edge having an edge type corresponding to the third attribute type, to a second node corresponding to the first term of the first attribute type and having a node type corresponding to the first attribute type.
4. The computer-implemented method of claim 1 further comprising:
providing access to a relationship database containing relationship distances of search results; and
determining whether the first, second, and third filters in the conversation state space have a relatedness measure exceeding a threshold value, wherein the threshold value is based on one or more of the relationship distances.
5. The computer-implemented method of claim 4 , wherein the relationship database comprises the relationship graph with nodes and edges.
6. The computer-implemented method of claim 5 , wherein a relationship distance between two entities is measured in a smallest number of hops between two nodes corresponding to the two entities.
7. The computer-implemented method of claim 6 , further comprising:
determining a measure of relatedness between the first speech input from a user and the second speech input from the user based on the relationship distance between a first entity corresponding to the first speech input and a second entity corresponding to the second speech input.
8. The computer-implemented method of claim 4 , further comprising determining a number of relationships that search results associated with the first, second, and third filters in the conversation state space have in the relationship database, wherein the threshold value is determined based on the number of relationships.
9. A system including a hardware processor for processing search requests, the system comprising:
input circuitry; and
control circuitry configured to:
receive, using the input circuitry, in a conversation state space defined by a first filter, the first filter comprising a first term of a first attribute type, a first speech input;
identify in the first speech input a second filter comprising a second term of a second attribute type;
generate a relationship graph that identifies relationships between a plurality of terms, including the first and second term, wherein the relationship graph is generated by the hardware processor retrieving and processing information from repositories to identify semantic associations between the plurality of terms;
select, based on the relationship graph, a first set of search results, wherein each search result of the first set of search results corresponds to a respective node of the relationship graph that is connected, by an edge having an edge type corresponding to the second attribute type, to a first node corresponding to the first term of the first attribute type and having a node type corresponding to the first attribute type;
generate for display the first set of search results;
receiving, using the input circuitry, in the conversation state space, a second speech input;
identify in the second speech input a third filter comprising a third term of a third attribute type;
determine, based on the relationship graph, whether any of the first search results match the first filter, the second filter, and the third filter;
in response to determining that no search results match: (a) the first filter and the second filter for which the first set of search results is being generated for display, and (b) the third filter which was received in the conversation state space defined by first filter and the second filter:
automatically remove the second filter from the conversation state space and replace it in the conversation state space with the third filter;
select a second set of search results based on the first filter and the third filter, and not the second filter; and
generate for display the second set of search results.
10. The system of claim 9 , wherein the control circuitry configured to determine, based on the relationship graph, whether any of the first search results match the first filter, the second filter, and the third filter is further configured to:
determine, for each search result of the first set of search results, whether a first node of the relationship graph corresponding to a respective search result is connected to a second node of the relationship graph having a node type corresponding to the third term by an edge having an edge type corresponding to the third attribute type.
11. The system of claim 9 , wherein the control circuitry configured to select the second set of search results based on the first filter and the third filter comprises:
select, based on the relationship graph, the second set of search results, wherein each search result of the second set of search results corresponds to a respective node of the relationship graph that is connected, by an edge having an edge type corresponding to the third attribute type, to a second node corresponding to the first term of the first attribute type and having a node type corresponding to the first attribute type.
12. The system of claim 9 , wherein the control circuitry is further configured to:
provide access to a relationship database containing relationship distances of search results; and
determine whether the first, second, and third filters in the conversation state space have a relatedness measure exceeding a threshold value, wherein the threshold value is based on one or more of the relationship distances.
13. The system of claim 12 , wherein the relationship database comprises the relationship graph with nodes and edges, each node representing an entity and each edge representing a direct relationship between two entities.
14. The system of claim 13 , wherein a relationship distance between two entities is measured in a smallest number of hops between two nodes corresponding to the two entities.
15. The system of claim 14 , wherein the control circuitry is further configured to:
determine a measure of relatedness between the first speech input from a user and the second speech input from the user based on the relationship distance between a first entity corresponding to the first speech input and a second entity corresponding to the second speech input.
16. The system of claim 12 , wherein the control circuitry is further configured to determine a number of relationships that search results associated with the first, second, and third filters in the conversation state space have in the relationship database, wherein the threshold value is determined based on the number of relationships.