CN104731923A - Construction method for Internet product review excavation noumenon lexicon - Google Patents
Construction method for Internet product review excavation noumenon lexicon Download PDFInfo
- Publication number
- CN104731923A CN104731923A CN201510138097.7A CN201510138097A CN104731923A CN 104731923 A CN104731923 A CN 104731923A CN 201510138097 A CN201510138097 A CN 201510138097A CN 104731923 A CN104731923 A CN 104731923A
- Authority
- CN
- China
- Prior art keywords
- word
- body dictionary
- comment
- dictionary
- commodity
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Abstract
The invention provides a construction method for an Internet product review excavation noumenon lexicon. The method comprises the steps that 1, attribute word noumenon lexicons are constructed: product reviews are acquired, and nouns are extracted using a word classification method and a part-of-speech tagging method according to product categories to form the attribute word noumenon lexicons; 2, an evaluation word noumenon lexicon is constructed; 3, a negative word noumenon lexicon is constructed: negative words are collected to construct the negative word noumenon lexicon; 4, a matched emotional word noumenon lexicon is constructed: matched feature words in the reviews are matched with corresponding matched emotional words according to the different kinds of product reviews based on the categories on the Internet to construct the matched emotional word noumenon lexicon; 5, a degree adverb noumenon lexicon is constructed: degree adverbs are collected for modifying the emotional words, and intensity levels and intensity values are given to the degree adverbs; 6, a stop word noumenon lexicon is constructed. According to the construction method for the Internet product review excavation noumenon lexicon, the query efficiency and the hit rate can be effectively promoted.
Description
Technical field
The present invention relates to internet comment on commodity, the construction method of body dictionary is excavated in the comment on commodity of especially a kind of internet.
Background technology
The global interconnection network data display of 2011, by the end of in Dec, 2011, global website quantity sum has reached 5.55 hundred million, and the number of global netizen has exceeded 2,000,000,000.The raising of internet popularity has driven the development of this network activity of ecommerce, increasing network messaging to be a mass of our network platform, and comment on commodity on shopping website is particularly evident.
2012, " two 11 " Alipay turnover on the same day realized rapidly increasing, and reaches 19,100,000,000 yuan, comprising 13,200,000,000 yuan, cat store, sky, and Taobao 5,900,000,000 yuan, order numbers reaches 1.058 hundred million; 2013, November 11 Alibaba's platform total turnover 350.19 hundred million; 2014, November 11 Alibaba's total business volume 571.1 hundred million, order total amount 2.79 hundred million.(above data are from official of Alibaba microblogging) Jingdone district official microblogging is announced, " two 11 " period three sky (on November 10 to 12) sales volume 2,500,000,000 yuan in 2014, three days order total amounts are single more than 6,800,000, be the same day on November 11st, 2012 order volume more than 3 times.
Except this two household appliances manufacturer, domestic also have much large-scale electric business as Dangdang.com etc.Add other more than 300 days the electric business's operation datas of shopping online outside two 11, statement of facts e-commerce initiative is more and more frequent, and produces the comment on commodity of magnanimity thus.
The consumption choice of consumer can be subject to the impact of the information exchanged between consumer, in the past, people often think that the suggestion heard from relative or friend there is information before most important consumption, now, before development in the technology more than ten years of being applied in over of internet makes people obtain consumption on network, the source of information is no longer confined to the relatives and friends of oneself, but has expanded the comment on commodity on shopping website to.In fact, these comments become the important sources that user obtains merchandise news.In traditional solid shop, client carefully can check quality and the quality of commodity, and the commodity in on-line shop, client can't see material object, will inevitably worry picture and gap in kind.Therefore for shopping at network, the related commentary of commodity is just extremely important, and these comments can help client to understand service and the public praise of commodity, help them to make correct decision-making, chooses oneself satisfied commodity.
In addition the producer and seller of these commodity also can from feedack income to some extent, improve and produce and service, improve the quality of commodity and the popularity of brand, potential consumer can also be excavated.
By observing the e-commerce website of existing main flow in a large number, wherein comment on commodity has following features:
1., in comment on commodity, most clause comments on for an attribute.Such as have comment " mobile phone is in one's hands has used a day, does manual work good, and software and game running are all good, and mobile phone is certified products, and the pocket-handkerchief given is also all good ", in the words, the object that the different clauses split by comma comment on is different.Minority ground, be distributed in different clauses, but these clauses is adjacent to the comment of an attribute.
2. in comment on commodity text, to the comment of same item attribute with have obvious boundary to the comment of other attributes.We are by a comment text, and each part split by punctuate or blank character is called clause.The content that most clause comprises is that an evaluation object (item property) adds that is evaluated a word.As " doing manual work good ".Also have part clause not have evaluation object, only have evaluation word, this kind of situation can use the evaluation object of acquiescence, as comment clause " just find after having used very very well ", can be understood as " quality is fine ".
3. comprise multiple evaluation attributes in some clause, as " mobile phone screen color is full ".
4. some clause is not the comment for commodity itself, if " wholehearted suggestion your company not with flexible cooperation, to this express company without language " is the comment to seller and logistics service quality.
5., there is relation of inclusion between the object that buyers comment in pair same commodity.Such as to a concrete mobile phone, in some comment, attribute word is " screen ", and in some comment, attribute word is " resolution ".
Day by day the comment on commodity increased sharply this allow people be difficult to read one by one.Affective tag is made up of evaluation object and evaluation word, contain the details of user comment, effectively can embody the core content of user comment, realize the conclusion of the information on commodity comment of internet electronic business website, retain as much as possible originally for effective content that the magnanimity of these commodity is commented on, facilitate again the quick grasp of comment reader.
The extraction of affective tag is the hot issue of text mining in recent years, previously there is excessive quantity research to the affective tag abstracting method based on dictionary, but rarely have the research for existing electric business's comment on commodity, previous method is used to process these comments, its efficiency is not high, mainly contain two reasons: one is that the dictionary content that uses is too wide in range, and to the vocabulary of current comment on commodity comprise spend completely low, when this just causes inquiry efficiency and hit rate low; Two is when text representation, research is before mostly it is considered that process large-scale document, and current Chinese comment on commodity text is all shorter, when being expressed as vector space model, dimension is very large, a lot of element is invalid Filling power, packing density is low, no matter is that storage or search efficiency are low.
The structure of body dictionary is the important step of the affective tag extractive technique based on dictionary.
Summary of the invention
The object of the invention is to, for the feature of current mainstream electronic commerce web site commodity comment, provide a kind of comment on commodity to excavate the construction method of body dictionary.The technical solution used in the present invention is:
A construction method for body dictionary is excavated in internet comment on commodity, comprises the steps:
Step 1, the structure of attribute word body dictionary: obtain comment on commodity, according to merchandise classification, utilizes segmenting method and part-of-speech tagging method to extract noun, forms attribute word body dictionary;
Step 2, evaluates the structure of word body dictionary: build commendatory term body dictionary and derogatory term body dictionary respectively;
Step 3, the structure of negative word body dictionary: collect negative word, builds negative word body dictionary;
Step 4, the structure of collocation emotion word body dictionary: according to the various comment on commodity based on classification on the net, corresponding collocation emotion word is mixed to the collocation Feature Words in comment, thus build collocation emotion word body dictionary;
Step 5, the structure of degree adverb body dictionary: collect degree adverb, described degree adverb for modifying emotion word, and gives intensity rank and intensity level to each degree adverb;
Step 6, the structure of stop words body dictionary: participle is carried out to the comment on commodity obtained, every bar comment is calculated to the characteristic frequency TF of each word, TF is the frequency that word occurs in comment text, selects the word that TF is high; For each word, calculate document frequency DF, DF is the ratio that comment text concentrates text number containing feature word and total textual data, selects the word that DF is high, in these words, manually selects stop words.
Further, in step 1, the attribute word body dictionary of formation has hierarchy; In attribute word body dictionary, the record format of entry is: (word, father node, class center word).
Further, in step 1, the word in attribute word body dictionary also has synonym, and each synonym part of speech at same level, and has a centre word.
Further, in step 2, concrete according to knowing the sentiment analysis word collection that net HowNet issues, front evaluation word wherein and positive emotion word are revised, then adds commendatory term body dictionary; To unfavorable ratings word wherein and negative emotion word, then add derogatory term body dictionary, build and evaluate word body dictionary.
Further, in step 2, evaluate word body dictionary also according to synonym classification, each class has class center word.
Further, in step 3, negative word body dictionary comprises following negative word: be not, or not need not, need not, never, not, do not have, do not have, don't, not, may not, not, not, not, not, no, deny, without, non-, be not, lose, exempt from, lack, prohibit, avoid, guard against, prevent, can't see.
Use the dictionary that current existing body dictionary construction method constructs, not high to efficiency during comment on commodity text-processing on existing mainstream electronic commerce website, mainly contain two reasons: one is that dictionary content is too wide in range, the body dictionary existed does not design for Chinese e-commerce website comment on commodity process specially, low in inquiry dictionary timeliness rate; Two is not enough to the vocabulary including degree needed for current comment on commodity process, and when this just causes inquiring about, hit rate is low.For above-mentioned two problems, the body dictionary in the present invention designs for existing Chinese e-commerce website comment on commodity process, can effectively promote search efficiency and hit rate.
Accompanying drawing explanation
Fig. 1 is the formation schematic diagram of body dictionary of the present invention.
Fig. 2 is the hierarchy schematic diagram of the attribute word body dictionary that the present invention relates to.
Fig. 3 is synonym part of speech and the centre word exemplary plot of the attribute word that the present invention relates to.
Fig. 4 is the evaluation word and centre word exemplary plot thereof that the present invention relates to.
Fig. 5 is process flow diagram of the present invention.
Embodiment
Below in conjunction with concrete drawings and Examples, the invention will be further described.
The structure that Chinese text excavates body dictionary lacks unified research method and specification, and in different research purposes and different applications, structure and the construction method of body dictionary are all not quite similar.For the feature of domestic main flow electricity business website Chinese comment on commodity, the body dictionary that the present invention proposes mainly comprises six class dictionaries, respectively: item property dictionary, commodity evaluate dictionary, negative dictionary, collocation emotion dictionary, degree adverb dictionary, stop words dictionary.
The construction method of body dictionary of the present invention mainly comprises following six steps:
Step 1, the structure of attribute word body dictionary:
The attribute difference commented between different types of merchandize is comparatively large, as women's dress class, cell phone type, skin care category.Women's dress class often there will be the attributes such as " collar " " hood ", and cell phone type then there will not be these attributes.Therefore attribute dictionary divides according to merchandise classification.
Attribute word is excavated from the magnanimity comment on commodity of e-commerce platform popular at present.The comment on commodity obtained, according to merchandise classification, such as " cell phone type " " women's dress class " utilizes segmenting method and part-of-speech tagging method to extract noun, after by manual sorting, and add neologisms, form attribute word body dictionary.Attribute dictionary example is as follows:
Due to item property limited amount, therefore can easily item property and synonym thereof be obtained by the mode of manual sorting.What the segmenting method based on body dictionary was conventional has reverse maximum matching process: direction coupling from right to left; Or Forward Maximum Method method: direction coupling from left to right, what adopt in this step is Forward Maximum Method segmenting method.Adopt the part-of-speech tagging method of Corpus--based Method in this step: the method utilizing machine learning, the speech training information that the mark of Corpus--based Method uses some statistic algorithms or model or collects in language material, then apply these information in the part-of-speech tagging of language material to be measured.For given word string, use the language message obtained, calculate a certain word in given context environmental according to statistical model and there is the probability of a certain mark, then obtain the suitable mark of part of speech according to probability.
As shown in Figure 2, according to the Concept Hierarchies of reality and the feature of comment on commodity, attribute word body dictionary should have hierarchy.The upper layer node of level comprises its lower level of child nodes on conceptual dependency, and the lower level of child nodes of such as " mobile phone " can be " shell " " screen " " color " etc.; The lower level of child nodes of " screen " can be " color " " resolution " etc.As shown in Figure 3, the word in attribute word body dictionary also should have synonym, and each synonym part of speech at same level, and has a centre word.Centre word can represent this type of synonym, and the frequency of occurrences is higher in comment on commodity.Such as " outward appearance " is centre word, and " texture " " shell " " design " " style " is synonym.
In each attribute dictionary, the record format of entry is as follows: (word, father node, class center word).
Partial words in Fig. 2 is as shown in the table.
| Word | Upper level centre word | Class center word |
| Mobile phone | NULL | Mobile phone |
| Outward appearance | Mobile phone | Outward appearance |
| Material | Mobile phone | Outward appearance |
| Material * | Screen | Material * |
The identical word of different levels can use different codings to distinguish." material " and " material * " is the attribute word of two different levels, and " material " is next level of child nodes of Fig. 2 interior joint " mobile phone ", and " material * " is next level of child nodes of node " screen ".
Step 2, evaluate the structure of word body dictionary:
The front evaluation word known in the sentiment analysis word collection (beta version) that net (HowNet, its network address is http://www.keenage.com/) is issued and positive emotion word are revised, then adds commendatory term body dictionary; Similarly, revise negative emotion word and unfavorable ratings word, then add derogatory term body dictionary.Correction work refers to concentrates from word the word extracting mistake, and " combined type " and " compound " that such as unfavorable ratings word is concentrated should belong to neutral word, and " wonderful work " in positive emotion word should be negative in the network life.Some words can also be added in addition, such as add " so-so " in derogatory sense body dictionary, add " gourmet's luck " in commendation body dictionary; In addition can also add some network flow langs, such as add " top ", " fabbing ", " Kazakhstan skin " in commendation body dictionary, " thunder people ", " without language ", " cup " are in derogatory sense body dictionary.
To carry out tissue matrix dictionary by merchandise classification different from attribute word, evaluate word and other dictionaries can be general between different type of merchandize.
The intensity of Sentiment orientation do not distinguished in emotion word in commendation and derogatory sense body dictionary, and the Sentiment orientation value of derogatory term is-1, and the Sentiment orientation value of commendatory term is 1.
As Fig. 4, as attribute word body dictionary, evaluate word body dictionary also according to synonym classification, each class has class center word.As " without language ", " thunder people " can be referred to same class, class center word can be " bad ".Same evaluation word also can be grouped into different classification, but will make difference on coding, as " cup ", can be grouped into the class that centre word is " bad ", also can be grouped into the class that centre word is " damage ".
Step 3, the structure of negative word body dictionary: negative word body dictionary carries out reprocessing gained from knowing net (HowNet) after acquisition.By knowing in net that the justice found containing Negation is former and observing the normal negative word occurred in Internet comment, negative word body dictionary can be obtained after entering artificial arrangement again, collect temporarily following 28 negative words: be not, or not need not, need not, never, not, do not have, do not have, don't, not, may not, not, not, not, not, no, deny, nothing, non-, be not, lose, exempt from, lack, prohibit, avoid, guard against, prevent, can't see.Know that the vocabulary in net has two very important concepts: the senses of a dictionary entry and justice former, the senses of a dictionary entry describes the one of vocabulary implication, and because the implication of vocabulary in Chinese is very complicated, in different linguistic context, same vocabulary can have several different implication; Another one concept is that justice is former, justice is former be considered to know the most basic in net, be not easy to the minimum semantic unit split again, the senses of a dictionary entry is all the former composition of various justice.
Step 4, the structure of collocation emotion word body dictionary: according to the various comment on commodity based on classification on the net, corresponding collocation emotion word is mixed to the collocation Feature Words in comment, thus build collocation emotion word body dictionary;
Collocation emotion word is just used to decorative features word, to the vocabulary that Feature Words adds a supplementary explanation.Such as: " this part clothes is had higher rating on the net, and price is also high ".Collocation emotion word " height " is had in upper sentence, only analyze this " height " and cannot judge that text is inclined to, the collocation Feature Words must modified according to it judges Sentiment orientation, when judgement (is evaluated, high) Sentiment orientation of this collocation time, the tendency of text is commendation, and judges (price, high) Sentiment orientation of this collocation time, the tendency of text is derogatory sense.Emotion word during the different Feature Words of this modification with different tendency is referred to as collocation emotion word in this article.
Collocation emotion word body dictionary example is as following table:
Step 5, the structure of degree adverb body dictionary: collect degree adverb, described degree adverb for modifying emotion word, and gives intensity rank and intensity level to each degree adverb;
Degree adverb, is generally used for modification emotion word, certain restriction is played to the degree of emotion word, can have an impact to the Sentiment orientation of text.Can reorder according to mild degree, such as: a little, to compare, very, extremely etc.She is a zingy girl.(beautiful is emotion word, is very degree adverb).
Such as: this part clothes is good-looking; This part clothes is seen very well; This part clothes is the most nice.The commendation degree of this three word there occurs obvious change, increases progressively successively.
Degree adverb intensity rank classification and intensity level assignment as shown in the table.
Step 6, the structure of stop words body dictionary: participle is carried out to the comment on commodity obtained, every bar comment is calculated to the characteristic frequency TF of each word, TF is the frequency that word occurs in comment text, selects the word that TF is high; For each word, calculate document frequency DF, DF is the ratio that comment text concentrates text number containing feature word and total textual data, selects the word that DF is high, and in these words, (TF high high with DF) manually selects stop words.
Stop words, needs to be filtered, to mask without any meaning to the implication of document.In general, conjunction, article, preposition all belong to stop words.
The Chinese stop words summed up has:, he, you, I, one, once, no, not only, can not, not only, not only, not only, must not, with, with it, and, individual, individual, for, in order to, be, and even, , it, one of, before, afterwards, and so on, also, also be, , in, so, people, other, what, from, thus, , just, with, above, below, what, in addition, He Wei, its, one, in fact, several, almost, namely, even if, even if, again, and, and, can, passable, , respectively, each, everybody, separately, , , , breathe out, how, with, breathe out, , , which, which, where
Except conjunction, article and conjunction are except stop words, some verbs, adjective and adverbial word also may be stop words, and information retrieval system can arrange an inactive vocabulary for filtering stop words.
The above, it is only present pre-ferred embodiments, not technical scope of the present invention is imposed any restrictions, thus every above embodiment is done according to technical spirit of the present invention any trickle amendment, equivalent variations and modification, all still belong in the scope of technical solution of the present invention.
Claims (6)
1. a construction method for body dictionary is excavated in internet comment on commodity, it is characterized in that, comprises the steps:
Step 1, the structure of attribute word body dictionary: obtain comment on commodity, according to merchandise classification, utilizes segmenting method and part-of-speech tagging method to extract noun, forms attribute word body dictionary;
Step 2, evaluates the structure of word body dictionary: build commendatory term body dictionary and derogatory term body dictionary respectively;
Step 3, the structure of negative word body dictionary: collect negative word, builds negative word body dictionary;
Step 4, the structure of collocation emotion word body dictionary: according to the various comment on commodity based on classification on the net, corresponding collocation emotion word is mixed to the collocation Feature Words in comment, thus build collocation emotion word body dictionary;
Step 5, the structure of degree adverb body dictionary: collect degree adverb, described degree adverb for modifying emotion word, and gives intensity rank and intensity level to each degree adverb;
Step 6, the structure of stop words body dictionary: participle is carried out to the comment on commodity obtained, every bar comment is calculated to the characteristic frequency TF of each word, TF is the frequency that word occurs in comment text, selects the word that TF is high; For each word, calculate document frequency DF, DF is the ratio that comment text concentrates text number containing feature word and total textual data, selects the word that DF is high, in these words, manually selects stop words.
2. the construction method of body dictionary is excavated in internet as claimed in claim 1 comment on commodity, it is characterized in that:
In step 1, the attribute word body dictionary of formation has hierarchy; In attribute word body dictionary, the record format of entry is: (word, father node, class center word).
3. the construction method of body dictionary is excavated in internet as claimed in claim 1 comment on commodity, it is characterized in that:
In step 1, the word in attribute word body dictionary also has synonym, and each synonym part of speech at same level, and has a centre word.
4. the construction method of body dictionary is excavated in internet as claimed in claim 1 comment on commodity, it is characterized in that:
In step 2, concrete according to knowing the sentiment analysis word collection that net HowNet issues, front evaluation word wherein and positive emotion word are revised, then adds commendatory term body dictionary; To unfavorable ratings word wherein and negative emotion word, then add derogatory term body dictionary, build and evaluate word body dictionary.
5. the construction method of body dictionary is excavated in internet as claimed in claim 1 comment on commodity, it is characterized in that:
In step 2, evaluate word body dictionary also according to synonym classification, each class has class center word.
6. the construction method of body dictionary is excavated in internet as claimed in claim 1 comment on commodity, it is characterized in that:
In step 3, negative word body dictionary comprises following negative word: be not, or not need not, need not, never, not, do not have, do not have, don't, not, may not, not, not, not, not, no, deny, without, non-, be not, lose, exempt from, lack, prohibit, avoid, guard against, prevent, can't see.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510138097.7A CN104731923A (en) | 2015-03-26 | 2015-03-26 | Construction method for Internet product review excavation noumenon lexicon |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510138097.7A CN104731923A (en) | 2015-03-26 | 2015-03-26 | Construction method for Internet product review excavation noumenon lexicon |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104731923A true CN104731923A (en) | 2015-06-24 |
Family
ID=53455810
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510138097.7A Pending CN104731923A (en) | 2015-03-26 | 2015-03-26 | Construction method for Internet product review excavation noumenon lexicon |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104731923A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105183847A (en) * | 2015-09-07 | 2015-12-23 | 北京京东尚科信息技术有限公司 | Feature information collecting method and device for web review data |
| CN106649260A (en) * | 2016-10-19 | 2017-05-10 | 中国计量大学 | Product feature structure tree construction method based on comment text mining |
| CN108108350A (en) * | 2017-11-29 | 2018-06-01 | 北京小米移动软件有限公司 | Name word recognition method and device |
| CN108153733A (en) * | 2017-12-26 | 2018-06-12 | 北京小度信息科技有限公司 | Comment on the sorting technique and device of quality |
| CN108399545A (en) * | 2017-02-06 | 2018-08-14 | 北京京东尚科信息技术有限公司 | E-commerce platform quality determining method and device |
| CN108596637A (en) * | 2018-04-24 | 2018-09-28 | 北京航空航天大学 | A kind of electric business service problem discovery system |
| CN108920448A (en) * | 2018-05-17 | 2018-11-30 | 南京大学 | A method of the comparison based on shot and long term memory network extracts |
| CN109145187A (en) * | 2018-07-23 | 2019-01-04 | 浙江大学 | Cross-platform electric business fraud detection method and system based on comment data |
| CN109190121A (en) * | 2018-09-03 | 2019-01-11 | 重庆工商大学 | Car review sentiment analysis method based on automobile body and part-of-speech rule |
| CN110322319A (en) * | 2019-06-26 | 2019-10-11 | 安徽景徽菜篮子电子商务有限公司 | A kind of electric business platform auto recommending method of user's evaluation |
| CN111651984A (en) * | 2019-02-19 | 2020-09-11 | 北京京东尚科信息技术有限公司 | Method and device for processing article description text and computer readable storage medium |
| CN112613612A (en) * | 2020-12-29 | 2021-04-06 | 合肥工业大学 | Method and device for constructing green design knowledge base based on patent library |
| CN112818682A (en) * | 2021-01-22 | 2021-05-18 | 深圳大学 | E-commerce data analysis method, equipment, device and computer-readable storage medium |
| CN114065769A (en) * | 2022-01-14 | 2022-02-18 | 四川大学 | Method, device, equipment and medium for training emotion reason pair extraction model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682074A (en) * | 2012-03-09 | 2012-09-19 | 浙江大学 | Product implicit attribute recognition method based on manifold learning |
| CN103678564A (en) * | 2013-12-09 | 2014-03-26 | 国家计算机网络与信息安全管理中心 | Internet product research system based on data mining |
| CN103778214A (en) * | 2014-01-16 | 2014-05-07 | 北京理工大学 | Commodity property clustering method based on user comments |
-
2015
- 2015-03-26 CN CN201510138097.7A patent/CN104731923A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102682074A (en) * | 2012-03-09 | 2012-09-19 | 浙江大学 | Product implicit attribute recognition method based on manifold learning |
| CN103678564A (en) * | 2013-12-09 | 2014-03-26 | 国家计算机网络与信息安全管理中心 | Internet product research system based on data mining |
| CN103778214A (en) * | 2014-01-16 | 2014-05-07 | 北京理工大学 | Commodity property clustering method based on user comments |
Non-Patent Citations (2)
| Title |
|---|
| 谈成访等: "基于语义分析的互联网产品评论挖掘", 《新乡学院学报》 * |
| 顾益军等: "中文停用词表的自动选取", 《北京理工大学学报》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105183847A (en) * | 2015-09-07 | 2015-12-23 | 北京京东尚科信息技术有限公司 | Feature information collecting method and device for web review data |
| CN106649260A (en) * | 2016-10-19 | 2017-05-10 | 中国计量大学 | Product feature structure tree construction method based on comment text mining |
| CN106649260B (en) * | 2016-10-19 | 2022-01-25 | 中国计量大学 | Product characteristic structure tree construction method based on comment text mining |
| CN108399545A (en) * | 2017-02-06 | 2018-08-14 | 北京京东尚科信息技术有限公司 | E-commerce platform quality determining method and device |
| CN108108350A (en) * | 2017-11-29 | 2018-06-01 | 北京小米移动软件有限公司 | Name word recognition method and device |
| CN108153733B (en) * | 2017-12-26 | 2021-07-09 | 北京星选科技有限公司 | Comment quality classification method and device |
| CN108153733A (en) * | 2017-12-26 | 2018-06-12 | 北京小度信息科技有限公司 | Comment on the sorting technique and device of quality |
| CN108596637A (en) * | 2018-04-24 | 2018-09-28 | 北京航空航天大学 | A kind of electric business service problem discovery system |
| CN108596637B (en) * | 2018-04-24 | 2022-05-06 | 北京航空航天大学 | Automatic E-commerce service problem discovery system |
| CN108920448A (en) * | 2018-05-17 | 2018-11-30 | 南京大学 | A method of the comparison based on shot and long term memory network extracts |
| CN108920448B (en) * | 2018-05-17 | 2021-09-14 | 南京大学 | Comparison relation extraction method based on long-term and short-term memory network |
| CN109145187A (en) * | 2018-07-23 | 2019-01-04 | 浙江大学 | Cross-platform electric business fraud detection method and system based on comment data |
| CN109190121A (en) * | 2018-09-03 | 2019-01-11 | 重庆工商大学 | Car review sentiment analysis method based on automobile body and part-of-speech rule |
| CN111651984A (en) * | 2019-02-19 | 2020-09-11 | 北京京东尚科信息技术有限公司 | Method and device for processing article description text and computer readable storage medium |
| CN110322319A (en) * | 2019-06-26 | 2019-10-11 | 安徽景徽菜篮子电子商务有限公司 | A kind of electric business platform auto recommending method of user's evaluation |
| CN112613612A (en) * | 2020-12-29 | 2021-04-06 | 合肥工业大学 | Method and device for constructing green design knowledge base based on patent library |
| CN112613612B (en) * | 2020-12-29 | 2022-08-02 | 合肥工业大学 | Method and device for constructing green design knowledge base based on patent library |
| CN112818682A (en) * | 2021-01-22 | 2021-05-18 | 深圳大学 | E-commerce data analysis method, equipment, device and computer-readable storage medium |
| CN114065769A (en) * | 2022-01-14 | 2022-02-18 | 四川大学 | Method, device, equipment and medium for training emotion reason pair extraction model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104731923A (en) | Construction method for Internet product review excavation noumenon lexicon | |
| CN104715049B (en) | Comment on commodity attribute word abstracting method based on body dictionary | |
| CN103729359B (en) | A kind of method and system recommending search word | |
| CN102831234B (en) | Personalized news recommendation device and method based on news content and theme feature | |
| CN102929873B (en) | Method and device for extracting searching value terms based on context search | |
| CN110175325A (en) | The comment and analysis method and Visual Intelligent Interface Model of word-based vector sum syntactic feature | |
| Zhong et al. | Using online reviews to explore consumer purchasing behaviour in different cultural settings | |
| CN103455487B (en) | The extracting method and device of a kind of search term | |
| CN104484431B (en) | A kind of multi-source Personalize News webpage recommending method based on domain body | |
| CN103886081A (en) | Information sending method and system | |
| CN102346761A (en) | Information processing device, related sentence providing method, and program | |
| CN104268230B (en) | A kind of Chinese micro-blog viewpoint detection method based on heterogeneous figure random walk | |
| CN104866496A (en) | Method and device for determining morpheme significance analysis model | |
| CN104035927A (en) | User behavior-based search method and system | |
| Zhang et al. | Combining sentiment analysis with a fuzzy kano model for product aspect preference recommendation | |
| CN105608166A (en) | Label extracting method and device | |
| CN104298732B (en) | The personalized text sequence of network-oriented user a kind of and recommendation method | |
| CN106909663A (en) | Based on tagging user Brang Preference behavior prediction method and its device | |
| Claster et al. | Naïve Bayes and unsupervised artificial neural nets for Cancun tourism social media data analysis | |
| CN103123624A (en) | Method of confirming head word, device of confirming head word, searching method and device | |
| CN108319376A (en) | A kind of input association recommendation method and device that optimization business word is promoted | |
| CN106547875A (en) | A kind of online incident detection method of the microblogging based on sentiment analysis and label | |
| Joshi et al. | Distributed word representations improve ner for e-commerce | |
| CN103150667A (en) | Individualization recommending method based on mechanical structure | |
| CN105468649A (en) | Method and apparatus for determining matching of to-be-displayed object |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150624 |
|
| RJ01 | Rejection of invention patent application after publication |