CN106156056A - A kind of Text Mode learning method and electronic equipment - Google Patents
A kind of Text Mode learning method and electronic equipment Download PDFInfo
- Publication number
- CN106156056A CN106156056A CN201510142414.2A CN201510142414A CN106156056A CN 106156056 A CN106156056 A CN 106156056A CN 201510142414 A CN201510142414 A CN 201510142414A CN 106156056 A CN106156056 A CN 106156056A
- Authority
- CN
- China
- Prior art keywords
- pattern
- text
- mode
- extension
- subpattern
- 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.)
- Granted
Links
Landscapes
- Machine Translation (AREA)
Abstract
The open a kind of Text Mode learning method of the present invention and electronic equipment.Described Text Mode learning method includes: obtain original text pattern；Described original text pattern is split as multiple first subpattern；Based on described first subpattern, from corpus, determine the first text collection；Learn described first text collection, obtain the first mode of extension.Use method or the electronic equipment of the present invention, it is possible to use existing language mode obtains more language mode from corpus learning, thus reduces time cost and human cost, and can sum up language mode more completely.
Description
Technical field
The present invention relates to text identification field, particularly relate to a kind of Text Mode learning method and electronics sets
Standby.
Background technology
With the development of information technology, the information that electronic equipment receives also gets more and more.Electronics sets
The standby information receiving increases so that the user of electronic equipment needs information content to be processed also to increase.
But, not all of information all needs user to process.A part of information can be from certain
Junk information is classified as in the degree of kind.For example, content is for " also worrying for invoice？My department handles state
The various invoices such as tax, land tax and value-added tax, Lee handles 13566666666 " note, generally classified
For refuse messages.
In order to save the time of user, the use feeling improving user is subject to, and in prior art, is showing information
Before showing to user, text recognition method is used to enter row filter, filtration to the information content.Text identification
Method mainly pre-sets some language modes being confirmed as refuse messages, by the information that receives with
The language mode pre-setting is compared, if comparison success, then this information is defined as refuse messages.
But, for the determination method of language mode of refuse messages in prior art, mainly artificial total
Knot.Artificial summary, needs manually to browse text, and row mode of going forward side by side arranges, therefore time cost and manpower
Relatively costly, and be difficult to sum up completely various patterns.
Content of the invention
It is an object of the invention to provide a kind of Text Mode learning method and electronic equipment, it is possible to use existing
Language mode obtains more language mode from corpus learning, thus reduces time cost and become with manpower
This, and language mode more completely can be summed up.
For achieving the above object, the invention provides following scheme:
A kind of Text Mode learning method, comprising:
Obtain original text pattern；
Described original text pattern is split as multiple first subpattern；
Based on described first subpattern, from corpus, determine the first text collection；
Learn described first text collection, obtain the first mode of extension.
Optionally, described from corpus, determine the first text collection, specifically include:
The first text collection matching with this first subpattern is determined from corpus.
Optionally, described first text collection of described study, after obtaining the first mode of extension, also includes:
Evaluate each described first mode of extension；
From described first mode of extension, determine the second mode of extension more than predetermined threshold value for the scoring.
Optionally, after described determination scoring is more than the second mode of extension of predetermined threshold value, also include:
Merge described original text pattern and described second mode of extension, the original text mould after being updated
Formula；
Original text pattern after described renewal is split as multiple second subpattern；
For any one the second subpattern, from corpus, determine second matching with this second subpattern
Text collection；
Learn described second text collection, obtain the 3rd mode of extension.
Optionally, described described original text pattern is split as multiple first subpattern, specifically includes:
Split out prefix pattern from described original text pattern；Described prefix pattern includes at least described initial literary composition
The initial character of this pattern；
Split out suffix pattern from described original text pattern；Described suffix pattern includes at least described initial literary composition
The last character of this pattern.
Optionally, described from corpus, determine the first text collection, specifically include:
With any one of first subpattern and any one of original text pattern phase while determination
The first text collection joined.
A kind of electronic equipment, comprising:
Original text pattern acquiring unit, is used for obtaining original text pattern；
First split cells, for being split as multiple first subpattern by described original text pattern；
First text collection determining unit, for based on described first subpattern, determines first from corpus
Text collection；
First unit, is used for learning described first text collection, obtains the first mode of extension.
Optionally, described first text collection determining unit, specifically includes:
First text collection determines subelement, matches for determination and this first subpattern from corpus
First text collection.
Optionally, also include:
Evaluation unit, at described first text collection of study, after obtaining the first mode of extension, evaluates
Each described first mode of extension；
Second mode of extension determining unit, for from described first mode of extension, determines that scoring is more than and presets
Second mode of extension of threshold value.
Optionally, also include:
Combining unit, for after determining the second mode of extension more than predetermined threshold value for the scoring, merges described
Original text pattern and described second mode of extension, the original text pattern after being updated；
Second split cells, for being split as multiple second submodule by the original text pattern after described renewal
Formula；
Second text collection determining unit, for for any one the second subpattern, determines from corpus
The second text collection matching with this second subpattern；
Second unit, is used for learning described second text collection, obtains the 3rd mode of extension.
Optionally, described first split cells, specifically includes:
Prefix pattern splits subelement, for splitting out prefix pattern from described original text pattern；Before described
Louver moudling formula is including at least the initial character of described original text pattern；
Suffix pattern splits subelement, for splitting out suffix pattern from described original text pattern；After described
Louver moudling formula is including at least the last character of described original text pattern.
Optionally, described first text collection determining unit, specifically includes:
Second text collection determines subelement, be used for determining simultaneously with any one of first subpattern and appointing
Anticipate the first text collection of a described original text patterns match.
The specific embodiment providing according to the present invention, the invention discloses techniques below effect:
Text Mode learning method in the embodiment of the present invention and electronic equipment, by by described original text mould
Formula is split as multiple first subpattern；Based on described first subpattern, from corpus, determine the first text set
Close；Learn described first text collection, obtain the first mode of extension；Can utilize existing language mode from
Corpus learning obtains more language mode, thus reduces time cost and human cost, and permissible
Sum up language mode more completely.
Brief description
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to enforcement
In example, the accompanying drawing of required use is briefly described, it should be apparent that, the accompanying drawing in describing below is only
Some embodiments of the present invention, for those of ordinary skill in the art, are not paying creative work
On the premise of, other accompanying drawing can also be obtained according to these accompanying drawings.
Fig. 1 is the flow chart of the Text Mode learning method embodiment 1 of the present invention；
Fig. 2 is the flow chart of the Text Mode learning method embodiment 2 of the present invention；
Fig. 3 is the flow chart of the Text Mode learning method embodiment 3 of the present invention；
Fig. 4 is the structure chart of the electronic equipment embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clearly
It Chu, is fully described by, it is clear that described embodiment is only a part of embodiment of the present invention, rather than
Whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not making creation
Property work under the premise of the every other embodiment that obtained, broadly fall into the scope of protection of the invention.
Understandable for enabling the above-mentioned purpose of the present invention, feature and advantage to become apparent from, below in conjunction with the accompanying drawings and
The present invention is further detailed explanation for detailed description of the invention.
The Text Mode learning method of the present invention, can apply to the electronic equipment with data processing function.
Described electronic equipment can be mobile phone, panel computer, desktop computer, server etc..
Fig. 1 is the flow chart of the Text Mode learning method embodiment 1 of the present invention.As it is shown in figure 1, the party
Method may include that
Step 101: obtain original text pattern；
Described original text pattern, it is also possible to be referred to as spermotype.Described original text pattern, is total in advance
The pattern that may be used for identifying certain certain types of text that knot obtains.
For example, an original text pattern is " I takes charge of offer invoice ".Can be to rubbish according to text pattern
Rubbish note is identified.Assume the phrase that there is " I takes charge of provides invoice " in a note, then can will be somebody's turn to do
Note is identified as refuse messages.
Step 102: described original text pattern is split as multiple first subpattern；
Described original text pattern can be split.For example, it is possible to split described original text pattern
For prefix pattern and suffix pattern.Assume to split " I takes charge of offer invoice " this original text pattern,
" I takes charge of offer " this prefix pattern, and " offer invoice " this suffix pattern at least can be provided.
Step 103: based on described first subpattern, determine the first text collection from corpus；
Described corpus, is the set of some statements, is by the material of Text Mode study.Assume needs
The Text Mode of learning spam note, then the statement in described corpus can be from the language in refuse messages
Sentence.
Described first subpattern can have multiple.For each the first subpattern, can from corpus really
Make the first text collection.Described first text collection can be the text matching with described first subpattern
Set, it is also possible to be the text set simultaneously mating with some first subpattern and some original text pattern
Close.
Step 104: learn described first text collection, obtain the first mode of extension.
Owing to described first subpattern has multiple, so described first text collection also can have multiple.For
Any one first text collection, can obtain new pattern from described first text collection learning.
For the learning process of described first text collection, mainly analyze the language in described first text collection
This same characteristic features is defined as new Text Mode by same characteristic features existing between Ju.
For ease of understanding the present invention, name the advantage so that said method to be described for the more specific example.
Still, as a example by " I takes charge of offer invoice " this original text pattern, " I takes charge of offer " is split out
This prefix pattern, and " offer invoice " this suffix pattern.In corpus, with " I takes charge of offer " this
The text (i.e. described first text collection) that one prefix pattern matches, comprising: " I takes charge of the various loan of offer
Business ", " I takes charge of the multiple financial business of offer ", " I takes charge of offer house lease business ".Then from front louver moudling
In the first text collection that formula matches, can learn to obtain " I takes charge of offer ... business " this pattern.
In corpus, the text (i.e. described first text collection) that matches with " offer invoice " this suffix pattern,
Including: " my department handles the miscellaneous services such as national tax, land tax and value-added tax, provides invoice ", " I takes charge of charg`e d'affaires and increase
Value tax, provide invoice ", " also for value-added tax worry？I takes charge of offer VAT invoice ".Then from rear
Sew in the first text collection of patterns match, can learn to obtain " ... value-added tax ... " this pattern.
In sum, in the present embodiment, by described original text pattern is split as multiple first subpattern；
Based on described first subpattern, from corpus, determine the first text collection；Learn described first text collection,
Obtain the first mode of extension；Existing language mode can be utilized to obtain more language from corpus learning
Pattern, thus reduce time cost and human cost, and language mode more completely can be summed up.
In actual application, in order to ensure study to mode of extension have higher for certain types of text
Accuracy of identification, described first text collection of described study, after obtaining the first mode of extension, can also include
Following step:
Evaluate each described first mode of extension；
From described first mode of extension, determine the second mode of extension more than predetermined threshold value for the scoring.
In above-mentioned steps, mark higher, represent that this first mode of extension is higher for the discrimination of text, know
Other result is more accurate.Mark lower, represent that this first mode of extension is lower for the discrimination of text, identify
Result is more inaccurate.
Described predetermined threshold value can be configured according to the actual requirements.When scoring is more than predetermined threshold value, then may be used
Higher for the discrimination of text to judge corresponding mode of extension.Therefore, above-mentioned steps can expand to first
Exhibition pattern enters row filter, therefrom filters out the higher mode of extension of discrimination.
Fig. 2 is the flow chart of the Text Mode learning method embodiment 2 of the present invention.As in figure 2 it is shown, the party
Method may include that
Step 201: obtain original text pattern；
Step 202: described original text pattern is split as multiple first subpattern；
Step 203: based on described first subpattern, determine the first text collection from corpus；
Step 204: learn described first text collection, obtain the first mode of extension.
Step 205: evaluate each described first mode of extension；
Step 206: from described first mode of extension, determines the second expanded mode more than predetermined threshold value for the scoring
Formula.
Step 207: merge described original text pattern and described second mode of extension, at the beginning of after being updated
Beginning Text Mode；
Described merge it can be understood as the merging of set in mathematics.Merge described original text pattern and institute
After stating the second mode of extension, in the original text pattern after the renewal obtaining, include described original text
All mode in pattern and described second mode of extension.
Step 208: the original text pattern after described renewal is split as multiple second subpattern；
Step 209: for any one the second subpattern, determines and this second subpattern phase from corpus
Second text collection of coupling；
Step 210: learn described second text collection, obtain the 3rd mode of extension.
Step 207, to 210, is the equal of step 201 to the circulation of step 204.In actual application, can
To repeat this circulation, until the original text pattern after Geng Xining no longer changes.As such, it is possible to really
Guarantor obtains Text Mode the most completely.
Additionally, for the mode of extension obtaining in each circulation, each described mode of extension all can be evaluated；
From described mode of extension, determine the step more than the mode of extension of predetermined threshold value for the scoring, filter out identification
The higher mode of extension of rate.
It should be noted that the described prefix pattern in the embodiment of the present invention includes at least described original text mould
The initial character of formula；Described suffix pattern is including at least the last character of described original text pattern.
Additionally, in the embodiment of the present invention, from corpus, determine the first text collection from described, specifically permissible
Including:
With any one of first subpattern and any one of original text pattern phase while determination
The first text collection joined.
Each step in the embodiment of the present invention can be mutually combined use, obtains various different realization side
Formula.Such as the following examples.
Fig. 3 is the flow chart of the Text Mode learning method embodiment 3 of the present invention.As it is shown on figure 3, the party
Method may include that
Step 301: original text pattern is split into multiple prefix pattern and suffix pattern；
Step 302: for any one prefix pattern, determine the first text with this prefix pattern match
Set；
Owing to each prefix pattern has corresponding first text collection, and prefix pattern can have many
Individual.Therefore multiple first text collection can be obtained.
Step 303: for any one suffix pattern, determine the second text mating with this suffix pattern
Set.
With step 302 in like manner, multiple second text collection can be obtained.
Step 304: from the first text collection learning to the first mode of extension；
Step 305: from the second text collection learning to the second mode of extension；
Step 306: determine simultaneously with in any one pattern in front suffix pattern and original text pattern
The 3rd text collection of any one patterns match；
Step 307: from the 3rd text collection learning to the 3rd mode of extension；
Step 308: merge described first mode of extension, the second mode of extension and the 3rd mode of extension,
To mode of extension set；
Step 309: evaluate each pattern in mode of extension set, obtain higher mode of extension of marking；
Step 310: mode of extension higher with scoring for current original text pattern is merged, obtains more
Original text pattern after Xin.
Repeat the above steps, until the original text pattern after Geng Xining no longer changes.
The present embodiment is relative to above-mentioned two embodiment, due to the increasing number of the mode of extension of merging, so
Can learn to obtain more mode of extension.
The invention also discloses a kind of electronic equipment.Described electronic equipment can have data processing function
Electronic equipment.For example, described electronic equipment can be mobile phone, panel computer, desktop computer, server etc..
Fig. 4 is the structure chart of the electronic equipment embodiment of the present invention.As shown in Figure 4, this electronic equipment is permissible
Including:
Original text pattern acquiring unit 401, is used for obtaining original text pattern；
First split cells 402, for being split as multiple first subpattern by described original text pattern；
First text collection determining unit 403, for based on described first subpattern, determines from corpus
First text collection；
First unit 404, is used for learning described first text collection, obtains the first mode of extension.
In the present embodiment, by described original text pattern is split as multiple first subpattern；Based on described
First subpattern, determines the first text collection from corpus；Learn described first text collection, obtain
One mode of extension；Existing language mode can be utilized to obtain more language mode from corpus learning,
Thus reduce time cost and human cost, and language mode more completely can be summed up.
In actual application, described first text collection determining unit 403, specifically may include that
First text collection determines subelement, matches for determination and this first subpattern from corpus
First text collection.
In actual application, this electronic equipment can also include:
Evaluation unit, at described first text collection of study, after obtaining the first mode of extension, evaluates
Each described first mode of extension；
Second mode of extension determining unit, for from described first mode of extension, determines that scoring is more than and presets
Second mode of extension of threshold value.
In actual application, this electronic equipment can also include:
Combining unit, for after determining the second mode of extension more than predetermined threshold value for the scoring, merges described
Original text pattern and described second mode of extension, the original text pattern after being updated；
Second split cells, for being split as multiple second submodule by the original text pattern after described renewal
Formula；
Second text collection determining unit, for for any one the second subpattern, determines from corpus
The second text collection matching with this second subpattern；
Second unit, is used for learning described second text collection, obtains the 3rd mode of extension.
In actual application, described first split cells 402, specifically may include that
Prefix pattern splits subelement, for splitting out prefix pattern from described original text pattern；Before described
Louver moudling formula is including at least the initial character of described original text pattern；
Suffix pattern splits subelement, for splitting out suffix pattern from described original text pattern；After described
Louver moudling formula is including at least the last character of described original text pattern.
In actual application, described first text collection determining unit 403, specifically may include that
Second text collection determines subelement, be used for determining simultaneously with any one of first subpattern and appointing
Anticipate the first text collection of a described original text patterns match.
In this specification, each embodiment uses the mode gone forward one by one to describe, and what each embodiment stressed is
With the difference of other embodiments, between each embodiment, identical similar portion sees mutually.For
For electronic equipment disclosed in embodiment, owing to it corresponds to the method disclosed in Example, so describe
Fairly simple, related part sees method part and illustrates.
Principle and embodiment to the present invention for the specific case used herein is set forth, above enforcement
The explanation of example is only intended to help to understand method and the core concept thereof of the present invention；Simultaneously for this area
Those skilled in the art, according to the thought of the present invention, all can change in specific embodiments and applications
Part.In sum, this specification content should not be construed as limitation of the present invention.
Claims (12)
1. a Text Mode learning method, it is characterised in that include:
Obtain original text pattern；
Described original text pattern is split as multiple first subpattern；
Based on described first subpattern, from corpus, determine the first text collection；
Learn described first text collection, obtain the first mode of extension.
2. method according to claim 1, it is characterised in that described determination first from corpus
Text collection, specifically includes:
The first text collection matching with this first subpattern is determined from corpus.
3. method according to claim 1, it is characterised in that described first text set of described study
Close, after obtaining the first mode of extension, also include:
Evaluate each described first mode of extension；
From described first mode of extension, determine the second mode of extension more than predetermined threshold value for the scoring.
4. method according to claim 3, it is characterised in that described determination scoring is more than default threshold
After second mode of extension of value, also include:
Merge described original text pattern and described second mode of extension, the original text mould after being updated
Formula；
Original text pattern after described renewal is split as multiple second subpattern；
For any one the second subpattern, from corpus, determine second matching with this second subpattern
Text collection；
Learn described second text collection, obtain the 3rd mode of extension.
5. method according to any one of claim 1 to 4, it is characterised in that described by described
Original text pattern is split as multiple first subpattern, specifically includes:
Split out prefix pattern from described original text pattern；Described prefix pattern includes at least described initial literary composition
The initial character of this pattern；
Split out suffix pattern from described original text pattern；Described suffix pattern includes at least described initial literary composition
The last character of this pattern.
6. method according to claim 1, it is characterised in that described determination first from corpus
Text collection, specifically includes:
With any one of first subpattern and any one of original text pattern phase while determination
The first text collection joined.
7. an electronic equipment, it is characterised in that include:
Original text pattern acquiring unit, is used for obtaining original text pattern；
First split cells, for being split as multiple first subpattern by described original text pattern；
First text collection determining unit, for based on described first subpattern, determines first from corpus
Text collection；
First unit, is used for learning described first text collection, obtains the first mode of extension.
8. electronic equipment according to claim 7, it is characterised in that described first text collection is true
Cell, specifically includes:
First text collection determines subelement, matches for determination and this first subpattern from corpus
First text collection.
9. electronic equipment according to claim 8, it is characterised in that also include:
Evaluation unit, at described first text collection of study, after obtaining the first mode of extension, evaluates
Each described first mode of extension；
Second mode of extension determining unit, for from described first mode of extension, determines that scoring is more than and presets
Second mode of extension of threshold value.
10. electronic equipment according to claim 9, it is characterised in that also include:
Combining unit, for after determining the second mode of extension more than predetermined threshold value for the scoring, merges described
Original text pattern and described second mode of extension, the original text pattern after being updated；
Second split cells, for being split as multiple second submodule by the original text pattern after described renewal
Formula；
Second text collection determining unit, for for any one the second subpattern, determines from corpus
The second text collection matching with this second subpattern；
Second unit, is used for learning described second text collection, obtains the 3rd mode of extension.
11. electronic equipments according to according to any one of claim 7 to 10, it is characterised in that described
First split cells, specifically includes:
Prefix pattern splits subelement, for splitting out prefix pattern from described original text pattern；Before described
Louver moudling formula is including at least the initial character of described original text pattern；
Suffix pattern splits subelement, for splitting out suffix pattern from described original text pattern；After described
Louver moudling formula is including at least the last character of described original text pattern.
12. electronic equipments according to claim 7, it is characterised in that described first text collection is true
Cell, specifically includes:
Second text collection determines subelement, be used for determining simultaneously with any one of first subpattern and appointing
Anticipate the first text collection of a described original text patterns match.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510142414.2A CN106156056B (en) | 2015-03-27 | 2015-03-27 | Text mode learning method and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510142414.2A CN106156056B (en) | 2015-03-27 | 2015-03-27 | Text mode learning method and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106156056A true CN106156056A (en) | 2016-11-23 |
| CN106156056B CN106156056B (en) | 2020-03-06 |
Family
ID=57339944
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510142414.2A Active CN106156056B (en) | 2015-03-27 | 2015-03-27 | Text mode learning method and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106156056B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451295A (en) * | 2017-08-17 | 2017-12-08 | 四川长虹电器股份有限公司 | A kind of method that deep learning training data is obtained based on grammer networks |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101055588A (en) * | 2007-05-25 | 2007-10-17 | 北京搜狗科技发展有限公司 | Method for catching limit word information, optimizing output and input method system |
| CN102236639A (en) * | 2010-04-28 | 2011-11-09 | 三星电子株式会社 | System and method for updating language model |
| CN103024746A (en) * | 2012-12-30 | 2013-04-03 | 清华大学 | System and method for processing spam short messages for telecommunication operator |
| CN103634473A (en) * | 2013-12-05 | 2014-03-12 | 南京理工大学连云港研究院 | Naive Bayesian classification based mobile phone spam short message filtering method and system |
| CN103813279A (en) * | 2012-11-14 | 2014-05-21 | ***通信集团设计院有限公司 | Junk short message detecting method and device |
| CN103870489A (en) * | 2012-12-13 | 2014-06-18 | 北京信息科技大学 | Chinese name self-extension recognition method based on search logs |
-
2015
- 2015-03-27 CN CN201510142414.2A patent/CN106156056B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101055588A (en) * | 2007-05-25 | 2007-10-17 | 北京搜狗科技发展有限公司 | Method for catching limit word information, optimizing output and input method system |
| CN102236639A (en) * | 2010-04-28 | 2011-11-09 | 三星电子株式会社 | System and method for updating language model |
| CN103813279A (en) * | 2012-11-14 | 2014-05-21 | ***通信集团设计院有限公司 | Junk short message detecting method and device |
| CN103870489A (en) * | 2012-12-13 | 2014-06-18 | 北京信息科技大学 | Chinese name self-extension recognition method based on search logs |
| CN103024746A (en) * | 2012-12-30 | 2013-04-03 | 清华大学 | System and method for processing spam short messages for telecommunication operator |
| CN103634473A (en) * | 2013-12-05 | 2014-03-12 | 南京理工大学连云港研究院 | Naive Bayesian classification based mobile phone spam short message filtering method and system |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451295A (en) * | 2017-08-17 | 2017-12-08 | 四川长虹电器股份有限公司 | A kind of method that deep learning training data is obtained based on grammer networks |
| CN107451295B (en) * | 2017-08-17 | 2020-06-30 | 四川长虹电器股份有限公司 | Method for obtaining deep learning training data based on grammar network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106156056B (en) | 2020-03-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107341220B (en) | Multi-source data fusion method and device | |
| CN108182245A (en) | The construction method and device of people's object properties classificating knowledge collection of illustrative plates | |
| CN105844424A (en) | Product quality problem discovery and risk assessment method based on network comments | |
| CN105224665A (en) | Wrong question management method and system | |
| CN101430714B (en) | Content structuring process method and system based on model | |
| CN104866308A (en) | Scenario image generation method and apparatus | |
| CN103473289A (en) | Device and method for completing communication addresses | |
| CN104933130A (en) | Comment information marking method and comment information marking device | |
| CN102968408A (en) | Method for identifying substance features of customer reviews | |
| CN104598577A (en) | Extraction method for webpage text | |
| CN103324622A (en) | Method and device for automatic generating of front page abstract | |
| CN105760474A (en) | Document collection feature word extracting method and system based on position information | |
| CN110427884A (en) | The recognition methods of the document structure of an article, device, equipment and storage medium | |
| CN105812554A (en) | Method and system for intelligently managing text messages in mobile phones | |
| CN102708164A (en) | Method and system for calculating movie expectation | |
| CN104317909A (en) | Method and device for verifying data of points of interest | |
| CN104462590A (en) | Information searching method and device | |
| CN103533169A (en) | Method for positioning and linking field of electronic business card based on mobile terminal | |
| US20100082625A1 (en) | Method for merging document clusters | |
| CN105095479A (en) | Mobile terminal and method for achieving photo classification management | |
| CN105138538A (en) | Cross-domain knowledge discovery-oriented topic mining method | |
| CN107391650B (en) | A kind of structuring method for splitting of document, apparatus and system | |
| CN104537459A (en) | BIM-based micro field system and method | |
| CN110417635B (en) | Automatic information reply method, device and equipment | |
| CN102063497A (en) | Open type knowledge sharing platform and entry processing method thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |