CN106657157B - Method for extracting session pair from session content - Google Patents
Method for extracting session pair from session content Download PDFInfo
- Publication number
- CN106657157B CN106657157B CN201710076035.7A CN201710076035A CN106657157B CN 106657157 B CN106657157 B CN 106657157B CN 201710076035 A CN201710076035 A CN 201710076035A CN 106657157 B CN106657157 B CN 106657157B
- Authority
- CN
- China
- Prior art keywords
- sentence
- type
- initiating
- conversation
- reply
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/1066—Session management
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/1066—Session management
- H04L65/1096—Supplementary features, e.g. call forwarding or call holding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/14—Session management
Abstract
The invention provides a method for extracting conversation pairs from conversation contents, which comprises the steps of acquiring the conversation contents, determining an initiating sentence and a replying sentence in the conversation contents according to the semantics of the conversation sentences in the conversation contents, and determining the types of the initiating sentence and the replying sentence according to a preset type judgment rule. In addition, aiming at some conversation sentences with complex or nonstandard structures, the embodiment of the invention can accurately extract conversation pairs with good integrity and high practicability.
Description
Technical Field
The invention relates to the technical field of communication, in particular to a method for extracting a session pair from session content.
Background
At present, a conversation pair or a question-answer pair extracted from conversation contents is often in a question-answer conversation pair form, and in an actual conversation process, conversation between two communicating parties does not completely conform to a question-answer conversation mode, for example, for a conversation sentence sent by a communicating party, the communicating party may reply several conversation sentences, or for a plurality of conversation sentences sent by the communicating party, the communicating party may reply only one conversation sentence.
Therefore, if the dialog pairs are extracted only in a question-and-answer manner, the following problems may occur:
(1) for some conversation contents which are not presented in a question-and-answer mode, the difficulty and the precision of extracting conversation pairs from the conversation contents are high. For example, for the conversation contents in the form of multiple initiating sentences and multiple replying sentences, when the conversation time pairs are extracted from the conversation contents, the replying sentences matched with each initiating sentence need to be analyzed, and the process is complex, the difficulty is high, and the precision is low.
(2) Because the existing question-answer pairs or conversation pairs extracted according to the conversation contents are generally relatively standard conversation sentences or conversation sentences with relatively simple structures, the conversation pairs with good integrity and high practicability cannot be accurately extracted aiming at some conversation sentences with complex or non-standard structures.
(3) In addition, the integrity of the conversation pair extracted in the form of a question-and-answer is easily damaged, so that the extracted conversation pair cannot accurately simulate a real conversation. In view of the above problems, the present invention provides a method for extracting a conversation pair from conversation contents according to the types of an initiating sentence and a replying sentence.
Disclosure of Invention
The invention provides a method for extracting a conversation pair from conversation contents, which aims to solve the technical problems of higher difficulty and lower precision in extracting the conversation pair in the prior art.
The invention provides a method for extracting conversation pairs from conversation contents, which comprises the following steps:
collecting conversation content;
determining an initiating sentence and a reply sentence in the conversation content according to the semantics of the conversation sentence in the conversation content;
judging whether the initiating sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type has the self upper and lower continuous conversation sentences or not, if so, further judging whether the initiating sentence can be combined with the self upper and lower continuous conversation sentences into a semantically associated sentence cluster, if so, extending the type of the initiating sentence into the sentence cluster initiating sentence type, otherwise, not extending the type;
extracting a basic conversation pair according to the initiating sentence and a reply sentence between the initiating sentence and the next initiating sentence;
and extracting at least one conversation pair according to the types of the initiating sentence and the replying sentence in the basic conversation pair and the basic conversation pair.
Further, determining the initiating sentence and the replying sentence in the conversation content according to the semantics of the conversation sentence in the conversation content comprises:
judging whether the conversation sentence in the conversation content has the text sent by the opposite communication party in a preset time interval, if not, determining the conversation sentence as an initiating sentence;
if yes, judging whether the conversation sentence is not semantically related to the text sent by the communication counterpart, if yes, determining the conversation sentence as an initiating sentence, otherwise, determining the conversation sentence as a recovering sentence.
Further, determining the type of the initiating sentence according to a preset type judgment rule includes:
judging whether the initiating sentence is a sentence with complete independent semantics, if so, judging whether the initiating sentence consists of a plurality of single sentences with complete independent semantics, if so, determining the type of the initiating sentence as a compound sentence initiating sentence type, otherwise, determining the type of the initiating sentence as a single sentence initiating sentence type; if not, judging whether the initiating sentence contains a single sentence with complete independent semantics, if so, determining the type of the initiating sentence as the type of the non-standard complex sentence initiating sentence, and if not, determining the type of the non-standard single sentence initiating sentence;
searching whether the initiating sentence of the non-standard single sentence initiating sentence type has the own continuous conversation sentences above and below, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard single sentence initiating sentence type can be combined with the own continuous conversation sentences above and below into a sentence with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard single sentence initiating sentence type into the non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
searching whether the initiating sentence of the non-standard compound sentence initiating sentence type has the own upper text and lower text continuous conversation sentences or not, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard compound sentence initiating sentence type can be combined with the own upper text and lower text continuous conversation sentences into sentences with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard compound sentence initiating sentence type into a non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
judging whether the starting sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type has the own continuous conversation sentence in the upper text and the lower text, if so, further judging whether the starting sentence can be combined with the own continuous conversation sentence in the upper text and the lower text to form a semantically related sentence cluster, if so, extending the type of the starting sentence to the sentence cluster starting sentence type in a derivative way, otherwise, not extending the derivative way.
Further, determining the type of the reply sentence according to a preset type judgment rule includes:
judging whether the reply sentence is a sentence with complete independent semantics, if so, judging whether the reply sentence is composed of a plurality of single sentences with complete independent semantics, if so, determining the type of the reply sentence as the type of the reply sentence, otherwise, determining the type of the reply sentence as the type of the single sentence; if not, judging whether the reply sentence contains a single sentence with complete independent semantics, if so, determining the type of the reply sentence as a non-standard reply sentence type, and if not, determining the type of the reply sentence as the non-standard single sentence reply sentence type;
searching whether a reply sentence of a non-standard single-sentence reply sentence type has a self upper text and lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard single-sentence reply sentence type can be combined with the self upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard single-sentence reply sentence type into a non-standard sentence cluster reply sentence type, and if not, performing no derivative expansion;
searching whether a reply sentence of a non-standard reply sentence type has a self upper text and lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard reply sentence type can be combined with the self upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard reply sentence type into a non-standard sentence group reply sentence type, and if not, performing no derivative expansion;
judging whether the reply sentences of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type have the self upper text and lower text continuous conversation sentences or not, if so, further judging whether the reply sentences can be combined with the self upper text and lower text continuous conversation sentences into semantically related sentence clusters, if so, extending the type of the reply sentences into the sentence cluster reply sentence type, otherwise, not extending the type of the reply sentences.
Further, extracting at least one conversation pair according to the base conversation pair, the type of the starting sentence in the base conversation pair, and the type of the reply sentence in the base conversation pair comprises:
performing derivative expansion on the types of the initiating sentences in the basic conversation pairs to obtain multiple types of initiating sentences;
carrying out derivative expansion on the types of the reply sentences in the basic conversation to obtain multiple types of reply sentences;
and combining at least one semantically related conversation pair for extraction according to the multiple types of initiating sentences and the multiple types of reply sentences.
Further, the collecting the session content includes:
the method comprises the steps of collecting conversation contents of an instant messaging account number, a mailbox account number, a microblog number and a mobile phone number of a user, wherein the conversation contents comprise conversation contents in text, picture, voice, video or cartoon formats.
The invention has the following beneficial effects:
the method for extracting the conversation pair from the conversation content, provided by the invention, comprises the steps of acquiring the conversation content, determining an initiating sentence and a replying sentence in the conversation content according to the semantics of the conversation sentence in the conversation content, determining the types of the initiating sentence and the replying sentence according to a preset type judgment rule, extracting a basic conversation pair according to the initiating sentence and the replying sentence between the initiating sentence and the next initiating sentence, and extracting at least one conversation pair according to the types of the initiating sentence and the replying sentence in the basic conversation pair and the basic conversation pair. In addition, aiming at some conversation sentences with complex or nonstandard structures, the embodiment of the invention can accurately extract the conversation pairs with good integrity and high practicability, thereby ensuring that the extracted conversation pairs can accurately simulate real conversations and the intelligent degree is higher. Furthermore, the conversation pair extracted by the embodiment of the invention has various forms, which is beneficial to accurately matching the intelligent reply content based on the conversation pair and obtaining the intelligent reply content with various forms by matching, and has higher practicability.
In addition to the objects, features and advantages described above, other objects, features and advantages of the present invention are also provided. The present invention will be described in further detail below with reference to the drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and are not to be construed as unduly limiting the invention. In the drawings:
FIG. 1 is a flow diagram of a method for extracting session pairs from session content in accordance with a preferred embodiment of the present invention;
FIG. 2 is a flow chart of a method for extracting session pairs from session content according to a simplified embodiment to which the preferred embodiment of the present invention is directed.
Detailed Description
The embodiments of the invention will be described in detail below with reference to the drawings, but the invention can be implemented in many different ways as defined and covered by the claims.
Referring to fig. 1, a preferred embodiment of the present invention provides a method for extracting session pairs from session content, including:
step S101, collecting conversation content;
step S102, determining an initiating sentence and a reply sentence in the conversation content according to the semantics of the conversation sentence in the conversation content;
step S103, according to a preset type judgment rule, determining the types of an initiating sentence and a reply sentence, wherein the types of the initiating sentence and the reply sentence comprise judging whether the initiating sentence of single sentence, complex sentence, nonstandard single sentence, nonstandard complex sentence and nonstandard sentence cluster types has an own upper text continuous conversation sentence and an own lower text continuous conversation sentence or not, if so, further judging whether the initiating sentence can be combined with the own upper text continuous conversation sentence and the own lower text continuous conversation sentence into a semantically associated sentence cluster, if so, deriving and expanding the type of the initiating sentence into a sentence cluster initiating sentence type, otherwise, not deriving and expanding;
step S104, extracting a basic conversation pair according to the initiating sentence and a reply sentence between the initiating sentence and the next initiating sentence;
and step S105, extracting at least one conversation pair according to the basic conversation pair and the types of the initiating sentence and the replying sentence in the basic conversation pair.
The method for extracting the conversation pair from the conversation content, provided by the embodiment of the invention, is characterized in that the conversation content is collected, the initiating sentence and the replying sentence in the conversation content are determined according to the semantics of the conversation sentence in the conversation content, the types of the initiating sentence and the replying sentence are determined according to the preset type judgment rule, the basic conversation pair is extracted according to the initiating sentence and the replying sentence between the initiating sentence and the next initiating sentence, and at least one conversation pair is extracted according to the types of the initiating sentence and the replying sentence in the basic conversation pair and the basic conversation pair. In addition, aiming at some conversation sentences with complex or nonstandard structures, the embodiment of the invention can accurately extract the conversation pairs with good integrity and high practicability, thereby ensuring that the extracted conversation pairs can accurately simulate real conversations and the intelligent degree is higher. Furthermore, the conversation pair extracted by the embodiment of the invention has various forms, which is beneficial to accurately matching the intelligent reply content based on the conversation pair and obtaining the intelligent reply content with various forms by matching, and has higher practicability.
It should be noted that, in this embodiment, before determining the types of the initiating sentence and the replying sentence, the types of the initiating sentence and the replying sentence and the type determination rule corresponding to the types are preset, so that the types of the initiating sentence and the replying sentence can be determined quickly according to the preset type determination rule. The initiating sentence in this embodiment specifically refers to a conversation sentence that is sent by a non-communication counterpart or a conversation sentence that is sent by a communication counterpart and has no semantic association with the above.
Optionally, determining the initiating sentence and the replying sentence in the conversation content according to the semantics of the conversation sentence in the conversation content includes:
judging whether the conversation sentence in the conversation content has the text sent by the opposite communication party in a preset time interval, if not, determining the conversation sentence as an initiating sentence;
if yes, judging whether the conversation sentence is not semantically related to the text sent by the communication counterpart, if yes, determining the conversation sentence as an initiating sentence, otherwise, determining the conversation sentence as a recovering sentence.
In order to accurately extract the conversation pair in the conversation content, the embodiment first determines the initiating sentence and the replying sentence in the conversation content according to the semantics of the conversation sentence in the conversation content, and then further determines the types of the initiating sentence and the replying sentence, so that the conversation pair is accurately extracted according to the types of the initiating sentence and the replying sentence. The specific process of determining the initiating sentence and the replying sentence in the conversation content according to the semantics of the conversation sentence in the conversation content comprises the steps of judging whether the conversation sentence in the conversation content has a text sent by a communication counterpart in a preset time interval, if not, determining the conversation sentence as the initiating sentence, if so, determining whether the conversation sentence is not semantically related to the text sent by the communication counterpart, if so, determining the conversation sentence as the initiating sentence, otherwise, determining the conversation sentence as the replying sentence.
In the actual conversation process, if there is no text sent by the communication counterpart in the preset time interval, the current conversation sentence is generally regarded as the starting sentence of the initiated conversation, that is, the initiating sentence. For example, if the current conversation sentence is a conversation sentence transmitted on day 3/12, the last conversation sentence is a conversation sentence transmitted on day 1/12 by the communication counterpart, and if the preset time interval is 1 day, it can be known through judgment that there is no text transmitted by the communication counterpart in the preset time interval in the current conversation sentence, the current conversation sentence is regarded as an initial sentence for initiating a conversation, that is, the current conversation sentence is judged as an initiation sentence. The preset time interval in this embodiment is specifically defined by a user, and may be, for example, 1 hour, half day, one month, and the like, that is, when it is determined that the current sentence is not the text sent by the other party in 1 hour, half day, one day, and one month, it is determined that the current sentence is the starting sentence.
In addition, when the conversation sentence has the text sent by the communication counterpart, the judgment can be made according to the actual conversation content, and the conversation sentence is possibly a reply sentence for replying the text sent by the communication counterpart; or may not reply the above sent by the communication counterpart, but re-initiate the starting sentence of the session; or the reply sentence sent by the communication counterpart and the initiation sentence for reinitiating the conversation are simultaneously replied. In this case, the present embodiment determines the type of the conversation sentence by determining whether the conversation sentence is semantically unrelated to the above text sent by the correspondent partner. It should be noted that, in this embodiment, whether the conversation sentence is semantically unrelated to the above text sent by the communication counterpart specifically means whether the conversation sentence includes a sentence semantically unrelated to the above text sent by the communication counterpart.
For example, when the conversation sentence has the text sent by the communication partner a, and the text sent by the communication partner a is "how recently? "then, aiming at the conversation sentence (the communication party B is" well "), the conversation sentence does not comprise the sentence which is not semantically related to the upper text sent by the communication party, and the conversation sentence is determined as a reply sentence; aiming at the conversation sentence in the second case (a communication party B:' do you pay telephone fee; for the conversation sentence in the third case (the communication party B: "is good, do you pay telephone fee.
In the embodiment, by judging whether the conversation sentence in the conversation content has the text sent by the communication counterpart in the preset time interval or not and judging whether the conversation sentence is semantically associated with the text sent by the communication counterpart or not when the conversation sentence in the conversation content has the text sent by the communication counterpart, the initiating sentence and the replying sentence in the conversation content can be accurately determined, and a foundation is laid for accurately extracting the conversation pair according to the determined initiating sentence and replying sentence subsequently.
Optionally, determining the type of the initiating sentence according to a preset type judgment rule includes:
judging whether the initiating sentence is a sentence with complete independent semantics, if so, judging whether the initiating sentence consists of a plurality of single sentences with complete independent semantics, if so, determining the type of the initiating sentence as a compound sentence initiating sentence type, otherwise, determining the type of the initiating sentence as a single sentence initiating sentence type; if not, judging whether the initiating sentence contains a single sentence with complete independent semantics, if so, determining the type of the initiating sentence as the type of the non-standard complex sentence initiating sentence, and if not, determining the type of the non-standard single sentence initiating sentence;
searching whether the initiating sentence of the non-standard single sentence initiating sentence type has the own continuous conversation sentences above and below, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard single sentence initiating sentence type can be combined with the own continuous conversation sentences above and below into a sentence with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard single sentence initiating sentence type into the non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
searching whether the initiating sentence of the non-standard compound sentence initiating sentence type has the own upper text and lower text continuous conversation sentences or not, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard compound sentence initiating sentence type can be combined with the own upper text and lower text continuous conversation sentences into sentences with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard compound sentence initiating sentence type into a non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
judging whether the starting sentences of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster types have the self upper and lower continuous conversation sentences or not, if so, further judging whether the starting sentences can be combined with the self upper and lower continuous conversation sentences into semantically related sentence clusters, if so, extending the type of the starting sentences of the determined types into the sentence cluster starting sentence types, otherwise, not extending the type of the starting sentences.
In actual implementation, the starting sentence may be presented in multiple types, such as a single sentence, a compound sentence, a non-standard sentence, and so on, and different types of starting sentences may affect or cause the extracted conversation pairs to be different. For this problem, the present embodiment determines the type of the initiating sentence according to a preset type determination rule. Specifically, firstly, on the premise that an initiating sentence has complete independent semantics, determining whether the initiating sentence is a single sentence or a compound sentence initiating sentence type by judging whether the initiating sentence is composed of one or more single sentences with complete independent semantics, and on the premise that the initiating sentence does not have complete independent semantics, determining whether the initiating sentence is a non-standard compound sentence or a non-standard single sentence initiating sentence type by judging whether the initiating sentence contains a single sentence with complete independent semantics; then, whether the type of the initiating sentence is derived and expanded into the type of the non-standard sentence group initiating sentence is determined by searching whether the initiating sentence of the type of the non-standard single sentence initiating sentence and the non-standard compound sentence initiating sentence has the own upper and lower continuous conversation sentences and whether the initiating sentence can be combined with the own upper and lower continuous conversation sentences into a sentence with complete independent semantics; and finally, determining whether the type of the initiating sentence can be derived and expanded into a sentence cluster initiating sentence type or not by judging whether the initiating sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type has own continuous conversation sentences above and below.
Specifically, the process of determining the type of the initiating sentence in this embodiment is essentially divided into three discrimination processes, that is, the first discrimination process is to discriminate each initiating sentence one by one according to four types of initiating sentences (single sentence, compound sentence, non-standard single sentence, and non-standard compound sentence); the second judging process is that after the first judging process is finished, judging whether the initiating sentences of the non-standard single sentence and non-standard compound sentence initiating sentence types can be further derived and expanded into non-standard sentence group initiating sentence types or not; the third judging process is to judge whether the starting sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type can be further derived and expanded into the sentence cluster starting sentence type after the second judging process is finished.
In the embodiment, the type of the initiating sentence is determined, so that on one hand, deep analysis of sentence structure and composition is favorably carried out on the initiating sentence, and on the other hand, based on type judgment and structural analysis carried out on the initiating sentence, more accurate extraction of conversation pairs with high practicability and various forms is favorably carried out. It should be noted that, in this embodiment, whether the initiating sentence has its own previous and following continuous conversational sentences specifically means whether the initiating sentence has a previous and following continuous conversational sentence sent by a sender sending the initiating sentence.
Optionally, determining the type of the reply sentence according to a preset type determination rule includes:
judging whether the reply sentence is a sentence with complete independent semantics, if so, judging whether the reply sentence is composed of a plurality of single sentences with complete independent semantics, if so, determining the type of the reply sentence as the type of the reply sentence, otherwise, determining the type of the reply sentence as the type of the single sentence; if not, judging whether the reply sentence contains a single sentence with complete independent semantics, if so, determining the type of the reply sentence as a non-standard reply sentence type, and if not, determining the type of the reply sentence as the non-standard single sentence reply sentence type;
searching whether a reply sentence of a non-standard single-sentence reply sentence type has a self upper text and lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard single-sentence reply sentence type can be combined with the self upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard single-sentence reply sentence type into a non-standard sentence cluster reply sentence type, and if not, performing no derivative expansion;
searching whether a reply sentence of a non-standard reply sentence type has a self upper text and lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard reply sentence type can be combined with the self upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard reply sentence type into a non-standard sentence group reply sentence type, and if not, performing no derivative expansion;
judging whether the reply sentences of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type have the own continuous conversation sentences above and below, if so, further judging whether the reply sentences can be combined with the own continuous conversation sentences above and below into semantically related sentence clusters, if so, extending the type of the reply sentences of the determined type into the sentence cluster reply sentence type, otherwise, not extending the type of the reply sentences.
The principle and process of judging the type of the reply sentence and the type of the initiating sentence are basically the same in the embodiment, so detailed description is omitted. In addition, in the embodiment, by determining the type of the reply sentence, on one hand, deep analysis of sentence structure and composition is favorably performed on the reply sentence, and on the other hand, based on type judgment and structural analysis performed on the reply sentence, more accurate extraction of conversation pairs with high practicability and various forms is favorably performed. It should be noted that, in this embodiment, whether a reply sentence has its own previous and following continuous conversational sentences specifically means whether the reply sentence has a previous and following continuous conversational sentence sent by a sender sending the reply sentence.
Optionally, extracting at least one conversation pair according to the base conversation pair, the type of the starting sentence in the base conversation pair, and the type of the recovering sentence in the base conversation pair includes:
deriving the types of the initiating sentences in the basic conversation pairs to obtain multiple types of initiating sentences;
deriving the types of the reply sentences in the basic conversation to obtain multiple types of reply sentences;
and combining at least one semantically related conversation pair for extraction according to the multiple types of initiating sentences and the multiple types of reply sentences.
Since the types of the initiating sentence and the replying sentence in this embodiment include a plurality of types, such as a single sentence, a compound sentence, a nonstandard single sentence, a nonstandard compound sentence, a nonstandard sentence group, a sentence group initiating sentence type, and a single sentence, a compound sentence, a nonstandard single sentence, a nonstandard compound sentence, a nonstandard sentence group reply sentence type, after the basic conversation pair is extracted, in order to more accurately extract conversation pairs with high practicability and various forms, in this embodiment, the type of the initiating sentence in the basic conversation pair is first derived and expanded to obtain a plurality of types of initiating sentences, then the type of the replying sentence in the basic conversation pair is derived and expanded to obtain a plurality of types of replying sentences, and finally, at least one semantically associated conversation pair is combined and extracted according to the plurality of types of initiating sentences and the plurality of replying sentences, so that a plurality of conversation pairs can be obtained by combination.
For example, assuming that the type of the initiating sentence is a complex sentence initiating sentence type and the type of the recovering sentence is a complex sentence recovering sentence type, after the type derivation and extension, a plurality of types of conversation pairs, such as a single sentence initiating sentence + a single sentence recovering sentence, a complex sentence initiating sentence + a single sentence recovering sentence, a single sentence initiating sentence + a complex sentence recovering sentence, a complex sentence initiating sentence + a complex sentence recovering sentence, and the like, can be extracted.
Optionally, the collecting the session content comprises:
the method comprises the steps of collecting conversation contents of an instant messaging account number, a mailbox account number, a microblog number and a mobile phone number of a user, wherein the conversation contents comprise conversation contents in text, picture, voice, video or cartoon formats.
The session content collected in this embodiment may be session content of an instant messaging account, a mailbox account, a microblog number, and a mobile phone number of the user, the session content includes session content in a text, picture, voice, video, or animation format, and when the session content is session content in a picture, voice, video, or animation format, the session content in a picture, voice, video, or animation format is first converted into session content in a text format.
The method for extracting session pairs from session content according to the present invention is further described below with respect to a simplified embodiment. Referring to fig. 2, a method for extracting session pairs from session contents according to a simplified embodiment of the present invention includes:
step S201, collecting session content.
Specifically, it is assumed that the session content collected in this embodiment is session content for a communication party a performing a session with an instant messaging account, a mailbox account, a microblog number, a mobile phone number, and a communication party B, where the session content is in a text, picture, voice, video, or animation format, and when the session content is in the voice, picture, video, or animation format, the session content in the voice, picture, video, or animation format is converted into the session content in the text format. For describing the process of extracting the session pair from the session content in detail, the present embodiment is described with the session content of the simple correspondent party a and the simple correspondent party B, which is as follows:
a: did you eat?
B: it is eaten.
B: do you wear?
A: help me collect money
A: is it charged?
B: the total of 100 yuan is paid.
B: there may be really many people in line.
Step S202, judging whether the conversation sentence in the conversation content has the text sent by the opposite communication party in a preset time interval, if not, determining the conversation sentence as an initiating sentence;
if yes, judging whether the conversation sentence is not semantically related to the text sent by the communication counterpart, if yes, determining the conversation sentence as an initiating sentence, otherwise, determining the conversation sentence as a recovering sentence.
Specifically, according to the above-mentioned judgment rule, the initiating sentence and the replying sentence in the conversation content can be determined, and it is assumed that the initiating sentence and the replying sentence in the conversation content are obtained by the judgment in table 1 in the present embodiment.
TABLE 1
| Initiating sentence | Replying sentence |
| Did you eat? | It is eaten. |
| Do you wear? | The total of 100 yuan is paid. |
| Help me collect money | There may be really many people in line. |
| Is it charged? |
Step S203, judging whether the initiating sentence is a sentence with complete independent semantics, if so, judging whether the initiating sentence is composed of a plurality of single sentences with complete independent semantics, if so, determining the type of the initiating sentence as a compound sentence initiating sentence type, otherwise, determining the type of the initiating sentence as a single sentence initiating sentence type, if not, judging whether the initiating sentence contains a single sentence with complete independent semantics, if so, determining the type of the initiating sentence as a non-standard compound sentence initiating sentence type, and if not, determining the type of the initiating sentence as a non-standard single sentence initiating sentence type;
searching whether the initiating sentence of the non-standard single sentence initiating sentence type has the own continuous conversation sentences above and below, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard single sentence initiating sentence type can be combined with the own continuous conversation sentences above and below into a sentence with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard single sentence initiating sentence type into the non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
searching whether the initiating sentence of the non-standard compound sentence initiating sentence type has the own upper text and lower text continuous conversation sentences or not, if not, not performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard compound sentence initiating sentence type can be combined with the own upper text and lower text continuous conversation sentences into sentences with complete independent semantics, if so, extending the type derivative of the initiating sentence of the non-standard compound sentence initiating sentence type into a non-standard sentence group initiating sentence type, and if not, not performing derivative expansion;
judging whether the starting sentences of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster types have the self upper and lower continuous conversation sentences or not, if so, further judging whether the starting sentences can be combined with the self upper and lower continuous conversation sentences into semantically related sentence clusters, if so, extending the type of the starting sentences of the determined types into the sentence cluster starting sentence types, otherwise, not extending the type of the starting sentences.
Specifically, it is assumed that the present embodiment first determines the type of the starting sentence according to the first determination process in step S203 as follows, which is specifically shown in table 2.
TABLE 2
| Serial number | Initiating sentence | Type (B) |
| First initiating sentence | Did you eat? | Single sentence |
| Second initiating sentence | Do you wear? | Single sentence |
| Third starting sentence | Help me collect money | Non-standard single sentence |
| The fourth initiating sentence | Is it charged? | Non-standard single sentence |
Then, according to the second determination procedure in step S203, it is determined whether the type of the non-standard single sentence and the non-standard compound sentence starting sentence is derived and expanded to the non-standard sentence group starting sentence type by determining whether the starting sentence of the non-standard single sentence and the non-standard compound sentence starting sentence type has its own upper and lower continuous conversation sentences and whether it can be merged with its own upper and lower continuous conversation sentences into a sentence with complete independent semantics. As can be seen from specific judgment, the third and fourth initiating sentences in this embodiment may be merged into a sentence with complete independent semantics, that is, the type derivation of the third and fourth initiating sentences may be extended to a non-standard sentence cluster initiating sentence type at this time, which is specifically shown in table 3.
TABLE 3
Finally, according to the third determination process in step S203, it is determined whether the starting sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence, and the nonstandard sentence group type can be further extended to the sentence group starting sentence type.
Specifically, as can be seen from table 3, the starting sentence cannot be further merged into the semantically related sentence group in this embodiment, that is, the starting sentence is not further subjected to derivative expansion in the last process. The type of the finally obtained initiating sentence is shown in table 3.
And step S204, determining the type of the reply sentence according to a preset type judgment rule.
The principle and process of determining the type of the reply sentence in the present embodiment are basically the same as the principle and process of determining the type of the initiating sentence, and therefore, detailed description is not given, and it is assumed that the embodiment specifically determines the type of the reply sentence as shown in table 5.
TABLE 5
Step S205, extracting a basic conversation pair according to the originating sentence plane and the reply sentence between the originating sentence and the next originating sentence.
Specifically, in this embodiment, for a first initiating sentence extraction session pair, it is first determined whether a replying sentence is present between the first initiating sentence and a next initiating sentence, if so, a basic session pair is extracted according to the initiating sentence and the replying sentence, and because a replying sentence is present between the first initiating sentence and the second initiating sentence, a basic session pair is extracted according to the first initiating sentence and the replying sentence. It should be noted that, in this embodiment, after determining that a reply sentence is included between an initiating sentence and a next initiating sentence, it is further required to calculate whether the initiating sentence and the reply sentence are semantically associated, and only under the condition that the semantics are associated, the basic conversation pair is extracted, otherwise, the basic conversation pair is not extracted. In this embodiment, assuming that the first initiating sentence and the first replying sentence are semantically associated, a basic session pair can be extracted, and assuming that the basic session pair is 1, the specific content of the basic session pair 1 is shown in table 6.
Similarly, in this embodiment, for the second initiating sentence, the basic session pair is extracted, and it is first determined whether there is a reply sentence between the second initiating sentence and the third initiating sentence, and if it is determined that there is no reply sentence between the second initiating sentence and the third initiating sentence, the second initiating sentence is discarded as the initiating sentence. Similarly, from the third and fourth starting sentences, it is assumed that the semantically associated base session pair 2 can be extracted, and the specific contents of the base session pair 2 are shown in table 6.
TABLE 6
And step S206, performing derivative expansion on the type of the initiating sentence in the basic conversation pair to obtain multiple types of initiating sentences.
Specifically, since the types of the initiating sentences in the present embodiment are six types, which are respectively a single sentence, a compound sentence, a nonstandard single sentence, a nonstandard compound sentence, a nonstandard sentence group, and a sentence group initiating sentence type, the present embodiment first performs derivative expansion according to the type of the initiating sentence in the basic session pair, and since the type of the initiating sentence in the basic session pair 1 in the present embodiment is a single sentence initiating sentence type, it cannot be further derivative expanded into other five initiating sentence types, only one type of initiating sentence, that is, the initiating sentence of a single sentence initiating sentence type, is included at this time, as shown in table 7. And according to the type of the starting sentence in the base session pair 2, the method can be further extended to other types of starting sentences, for example, a single sentence starting sentence type, as shown in table 7.
TABLE 7
And step S207, performing derivative expansion on the types of the reply sentences in the basic conversation to obtain multiple types of reply sentences.
Specifically, since the reply sentence types in the present embodiment are six types, which are respectively a single sentence, a compound sentence, a nonstandard single sentence, a nonstandard compound sentence, a nonstandard sentence group, and a sentence group reply sentence type. Therefore, the embodiment first performs derivative expansion on the type of the reply sentence in the base session pair according to table 8, and since the type of the reply sentence in the base session pair 1 in the embodiment is a single-sentence reply sentence type, it cannot be further derivative expanded into other five reply sentence types, so that only one type of reply sentence, i.e. the reply sentence of the single-sentence reply sentence type, is included at this time. And according to the type of the reply sentence in the base session pair 2, the method can be further extended to other types of reply sentences, for example, a type of the reply sentence, as shown in table 8.
TABLE 8
And step S208, combining at least one semantically related conversation pair for extraction according to the initiating sentences of various types and the reply sentences of various types.
Specifically, since only one type of the initiating sentence and the replying sentence is provided for the base conversation pair 1, only one conversation pair can be extracted at this time, and since the types of the initiating sentence and the replying sentence are multiple for the base conversation pair 2, multiple conversation pairs can be obtained by combining, specifically see table 9, where table 9 is 6 conversation pairs extracted from the base conversation pair 2.
TABLE 9
Therefore, according to the embodiment, a plurality of conversation pairs can be obtained according to the types of the initiating sentence and the replying sentence, so that the extracted conversation pairs are various in form and high in precision.
The method for extracting the conversation pair from the conversation content, provided by the embodiment of the invention, is characterized in that the conversation content is collected, the initiating sentence and the replying sentence in the conversation content are determined according to the semantics of the conversation sentence in the conversation content, the types of the initiating sentence and the replying sentence are determined according to the preset type judgment rule, the basic conversation pair is extracted according to the initiating sentence and the replying sentence between the initiating sentence and the next initiating sentence, and at least one conversation pair is extracted according to the types of the initiating sentence and the replying sentence in the basic conversation pair and the basic conversation pair. In addition, aiming at some conversation sentences with complex or nonstandard structures, the embodiment of the invention can accurately extract the conversation pairs with good integrity and high practicability, thereby ensuring that the extracted conversation pairs can accurately simulate real conversations and the intelligent degree is higher. Furthermore, the conversation pairs extracted by the embodiment of the invention have various forms, which is favorable for accurately matching intelligent reply contents based on the conversation pairs and obtaining the intelligent reply contents with various forms through matching, and the practicability is .
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. A method for extracting session pairs from session content, comprising:
collecting conversation content;
determining an initiating sentence and a reply sentence in the conversation content according to the semantics of the conversation sentence in the conversation content;
judging whether the initiating sentence of the single sentence type, the compound sentence type, the nonstandard single sentence type, the nonstandard compound sentence type and the nonstandard sentence cluster type has an own continuous conversation sentence with an upper text and a lower text, if so, further judging whether the initiating sentence can be combined with the own continuous conversation sentence with the upper text and the lower text into a semantically related sentence cluster, if so, deriving and expanding the type of the initiating sentence into a sentence cluster initiating sentence type, otherwise, not deriving and expanding;
extracting a basic conversation pair according to the initiating sentence and a reply sentence between the initiating sentence and the next initiating sentence; and extracting at least one conversation pair according to the types of the initiating sentence and the replying sentence in the basic conversation pair.
2. The method of claim 1, wherein determining the initiating sentence and the replying sentence in the conversational content according to the semantics of the conversational sentence in the conversational content comprises:
judging whether the conversation sentence in the conversation content has the text sent by the opposite communication party in a preset time interval, and if not, determining the conversation sentence as an initiating sentence;
if so, judging whether the conversation sentence is not semantically associated with the upper text sent by the communication counterpart, if so, determining the conversation sentence as an initiating sentence, otherwise, determining the conversation sentence as a recovering sentence.
3. The method of claim 2, wherein determining the type of the initiating sentence according to a preset type judgment rule comprises:
judging whether the initiating sentence is a sentence with complete independent semantics or not, if so, judging whether the initiating sentence consists of a plurality of single sentences with complete independent semantics or not, if so, determining the type of the initiating sentence as a compound sentence initiating sentence type, otherwise, determining the type of the initiating sentence as a single sentence initiating sentence type; if not, judging whether the initiating sentence contains a single sentence with complete independent semantics, if so, determining the type of the initiating sentence as a non-standard complex sentence initiating sentence type, and if not, determining the type of the initiating sentence as a non-standard single sentence initiating sentence type;
searching whether the initiating sentence of the non-standard single sentence initiating sentence type has an own upper text and a lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard single sentence initiating sentence type can be combined with the own upper text and lower text continuous conversation sentences into a sentence with complete independent semantics, if so, performing derivative expansion on the initiating sentence of the non-standard single sentence initiating sentence type into a non-standard sentence group initiating sentence type, and if not, performing no derivative expansion;
searching whether the initiating sentence of the non-standard compound sentence initiating sentence type has an own upper text and a lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the initiating sentence of the non-standard compound sentence initiating sentence type can be combined with the own upper text and lower text continuous conversation sentences into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the initiating sentence of the non-standard compound sentence initiating sentence type into a non-standard sentence group initiating sentence type, and if not, performing derivative expansion;
judging whether the initiating sentence of the single sentence, the compound sentence, the nonstandard single sentence, the nonstandard compound sentence and the nonstandard sentence cluster type has the self upper text and lower text continuous conversation sentences or not, if so, further judging whether the initiating sentence can be combined with the self upper text and lower text continuous conversation sentences into a semantically related sentence cluster, if so, deriving and expanding the type of the initiating sentence into the sentence cluster initiating sentence type, otherwise, not performing derivation and expansion.
4. The method of claim 2, wherein determining the type of the reply sentence according to a preset type determination rule comprises:
judging whether the reply sentence is a sentence with complete independent semantics, if so, judging whether the reply sentence is composed of a plurality of single sentences with complete independent semantics, if so, determining the type of the reply sentence as the type of the reply sentence, otherwise, determining the type of the reply sentence as the type of the single sentence; if not, judging whether the reply sentence contains a single sentence with complete independent semantics, if so, determining the type of the reply sentence as a non-standard reply sentence type, and if not, determining the type of the reply sentence as a non-standard single sentence reply sentence type;
searching whether the reply sentence of the non-standard single sentence reply sentence type has an own upper text and a lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard single sentence reply sentence type can be combined with the own upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard single sentence reply sentence type into a non-standard sentence group reply sentence type, and if not, performing derivative expansion;
searching whether the reply sentence of the non-standard reply sentence type has an own upper text and a lower text continuous conversation sentence, if not, performing derivative expansion, if so, further judging whether the reply sentence of the non-standard reply sentence type can be combined with the own upper text and lower text continuous conversation sentence into a sentence with complete independent semantics, if so, performing derivative expansion on the type of the reply sentence of the non-standard reply sentence type into a non-standard sentence group reply sentence type, and if not, performing no derivative expansion;
judging whether the reply sentences of the single sentence type, the compound sentence type, the nonstandard single sentence type, the nonstandard compound sentence type and the nonstandard sentence group type have own continuous conversation sentences of the upper text and the lower text, if so, further judging whether the reply sentences can be combined with the own continuous conversation sentences of the upper text and the lower text into semantically related sentence groups, if so, deriving and expanding the type of the reply sentences into the sentence group reply sentence type, otherwise, not performing derivation and expansion.
5. The method of claim 4, wherein extracting at least one conversation pair according to a base conversation pair, a type of sentence originating in the base conversation pair, and a type of sentence replying in the base conversation pair comprises:
performing derivative expansion on the type of the initiating sentence in the basic conversation pair to obtain multiple types of initiating sentences;
performing derivative expansion on the types of the reply sentences in the basic conversation to obtain multiple types of reply sentences; and combining at least one semantically related conversation pair for extraction according to the multiple types of the initiating sentences and the multiple types of the reply sentences.
6. The method of claim 5, wherein capturing session content comprises:
the method comprises the steps of collecting conversation contents of an instant messaging account number, a mailbox account number, a microblog number and a mobile phone number of a user, wherein the conversation contents comprise conversation contents in text, picture, voice, video or cartoon formats.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710076035.7A CN106657157B (en) | 2017-02-13 | 2017-02-13 | Method for extracting session pair from session content |
| PCT/CN2017/098456 WO2018145436A1 (en) | 2017-02-13 | 2017-08-22 | Method for extracting conversation pair from conversation content |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710076035.7A CN106657157B (en) | 2017-02-13 | 2017-02-13 | Method for extracting session pair from session content |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106657157A CN106657157A (en) | 2017-05-10 |
| CN106657157B true CN106657157B (en) | 2020-04-07 |
Family
ID=58844733
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710076035.7A Active CN106657157B (en) | 2017-02-13 | 2017-02-13 | Method for extracting session pair from session content |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN106657157B (en) |
| WO (1) | WO2018145436A1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106657157B (en) * | 2017-02-13 | 2020-04-07 | 长沙军鸽软件有限公司 | Method for extracting session pair from session content |
| CN107608946A (en) * | 2017-09-30 | 2018-01-19 | 努比亚技术有限公司 | Word key content extracting method and corresponding mobile terminal |
| CN111970311B (en) * | 2020-10-23 | 2021-02-02 | 北京世纪好未来教育科技有限公司 | Session segmentation method, electronic device and computer readable medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103412855A (en) * | 2013-06-27 | 2013-11-27 | 华中师范大学 | Method and system for automatic identification of relative words in complex sentence of modern Chinese language |
| CN103430578A (en) * | 2010-10-27 | 2013-12-04 | 诺基亚公司 | Method and apparatus for identifying conversation in multiple strings |
| CN104881402A (en) * | 2015-06-02 | 2015-09-02 | 北京京东尚科信息技术有限公司 | Method and device for analyzing semantic orientation of Chinese network topic comment text |
| CN105389296A (en) * | 2015-12-11 | 2016-03-09 | 小米科技有限责任公司 | Information partitioning method and apparatus |
| CN105528403A (en) * | 2015-12-02 | 2016-04-27 | 小米科技有限责任公司 | Target data identification method and apparatus |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8296127B2 (en) * | 2004-03-23 | 2012-10-23 | University Of Southern California | Discovery of parallel text portions in comparable collections of corpora and training using comparable texts |
| CN101599071B (en) * | 2009-07-10 | 2012-04-18 | 华中科技大学 | Automatic extraction method of conversation text topic |
| CN104166643A (en) * | 2014-08-19 | 2014-11-26 | 南京金娃娃软件科技有限公司 | Dialogue act analyzing method in intelligent question-answering system |
| CN106709072A (en) * | 2017-02-13 | 2017-05-24 | 长沙军鸽软件有限公司 | Method of obtaining intelligent conversation reply content based on shared corpora |
| CN106874452A (en) * | 2017-02-13 | 2017-06-20 | 长沙军鸽软件有限公司 | A kind of method for obtaining session reply content |
| CN106844735A (en) * | 2017-02-13 | 2017-06-13 | 长沙军鸽软件有限公司 | A kind of method of the personal exclusive corpus of automatic foundation |
| CN106844734B (en) * | 2017-02-13 | 2023-01-24 | 长沙军鸽软件有限公司 | Method for automatically generating session reply content |
| CN106649280B (en) * | 2017-02-13 | 2019-07-09 | 长沙军鸽软件有限公司 | A method of creating shared corpus |
| CN106874451A (en) * | 2017-02-13 | 2017-06-20 | 长沙军鸽软件有限公司 | A kind of method of the personal exclusive corpus of automatic foundation |
| CN106657157B (en) * | 2017-02-13 | 2020-04-07 | 长沙军鸽软件有限公司 | Method for extracting session pair from session content |
| CN106844347A (en) * | 2017-02-13 | 2017-06-13 | 长沙军鸽软件有限公司 | A kind of method that session pair is extracted according to session content |
| CN107015968A (en) * | 2017-04-27 | 2017-08-04 | 长沙军鸽软件有限公司 | A kind of method that session is actively initiated based on shared corpus |
-
2017
- 2017-02-13 CN CN201710076035.7A patent/CN106657157B/en active Active
- 2017-08-22 WO PCT/CN2017/098456 patent/WO2018145436A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103430578A (en) * | 2010-10-27 | 2013-12-04 | 诺基亚公司 | Method and apparatus for identifying conversation in multiple strings |
| CN103412855A (en) * | 2013-06-27 | 2013-11-27 | 华中师范大学 | Method and system for automatic identification of relative words in complex sentence of modern Chinese language |
| CN104881402A (en) * | 2015-06-02 | 2015-09-02 | 北京京东尚科信息技术有限公司 | Method and device for analyzing semantic orientation of Chinese network topic comment text |
| CN105528403A (en) * | 2015-12-02 | 2016-04-27 | 小米科技有限责任公司 | Target data identification method and apparatus |
| CN105389296A (en) * | 2015-12-11 | 2016-03-09 | 小米科技有限责任公司 | Information partitioning method and apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018145436A1 (en) | 2018-08-16 |
| CN106657157A (en) | 2017-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106657157B (en) | Method for extracting session pair from session content | |
| CN102118510B (en) | Contact correlation method, server and mobile terminal | |
| CN104715752A (en) | Voice recognition method, voice recognition device and voice recognition system | |
| EP2782369A1 (en) | Information prompt method and device and terminal equipment | |
| CN102761848B (en) | Method for determining short message intercepting key words | |
| CN105072238A (en) | Method and apparatus for creating contact list according to note information of newly-added number | |
| CN103501374A (en) | Telephone book sequencing method and device as well as terminal | |
| CN106656732A (en) | Scene information-based method and device for obtaining chat reply content | |
| CN106649404A (en) | Session scene database creation method and apparatus | |
| CN103167167A (en) | Mobile terminal and extraction method of communication contact person information | |
| CN106649410A (en) | Method and device for obtaining chitchat reply content | |
| CN102497391A (en) | Server, mobile terminal and prompt method | |
| CN106874452A (en) | A kind of method for obtaining session reply content | |
| US20150074254A1 (en) | Crowd-sourced clustering and association of user names | |
| CN103249034A (en) | Method and device for acquiring contact information | |
| CN104702759A (en) | Address list setting method and address list setting device | |
| CN105100353A (en) | Method for performing address book grouping on newly-added contact of mobile terminal | |
| CN106709072A (en) | Method of obtaining intelligent conversation reply content based on shared corpora | |
| CN106294792B (en) | The method for building up of correlation inquiry system and establish system | |
| CN103533169A (en) | Method for positioning and linking field of electronic business card based on mobile terminal | |
| CN106844734B (en) | Method for automatically generating session reply content | |
| CN108540677A (en) | Method of speech processing and system | |
| CN106874451A (en) | A kind of method of the personal exclusive corpus of automatic foundation | |
| CN105025489A (en) | Method for automatically shielding junk short messages | |
| CN107015968A (en) | A kind of method that session is actively initiated based on shared corpus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |