CN100444167C - Method for managing and searching dictionary with perfect even numbers group TRIE Tree - Google Patents
Method for managing and searching dictionary with perfect even numbers group TRIE Tree Download PDFInfo
- Publication number
- CN100444167C CN100444167C CNB2005101306903A CN200510130690A CN100444167C CN 100444167 C CN100444167 C CN 100444167C CN B2005101306903 A CNB2005101306903 A CN B2005101306903A CN 200510130690 A CN200510130690 A CN 200510130690A CN 100444167 C CN100444167 C CN 100444167C
- Authority
- CN
- China
- Prior art keywords
- dictionary
- trie tree
- array
- node
- value
- 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
Abstract
The present invention relates to the field of natural language processing and the field of information search, particularly to a perfect double array TRIE tree dictionary managing and searching method. The Trie tree structure is converted into two linear arrays to be expressed. An optimizing policy and a self-adapting coding solution to automatically code to characters by taking a byte as a coding unit are provided. The procedure comprises that: (1) a dictionary is expressed by the Trie tree structure; (2) the Trie tree is converted into two linear arrays; (3) according to a user's input content, the dictionary is searched by making use of the two generated arrays. The perfect double array TRIE tree dictionary managing and searching method comprises the following specific steps that: step 1, the dictionary is automatically coded by taking a byte as a unit so as to generate a sequence code file; setp 2, the dictionary is expressed with the Trie tree which is converted into two arrays to be expressed by making use of the sequence code file; step 3, words submitted by a user are searched in the two generated linear arrays.
Description
Technical field
The present invention relates to natural language processing field and information retrieval field, particularly a kind of dictionary with perfect even numbers group TRIE Tree management and search method.
Background technology
For the storage and the search of mass data, all adopt index structure to realize at present usually.General index structure commonly used comprises linear directory table, inverted list, hash (Hash) table and various search tree.
Linear directory is a kind of index structure of static state, is unfavorable for upgrading, whenever making the position that need change each index entry in the concordance list when upgrading successively.Inverted list also is a static index, and it is the same with the linear directory table, all can only sequential search or search by half when searching for wherein data.
Hash (Hash) method then is to set up a definite respective function to concern Hash () between the memory location of list item and its key, makes each key corresponding with a memory location in the structure.Only need carry out function calculation to the key of list item during search, the functional value of trying to achieve promptly is the memory location of list item, has avoided repeatedly key relatively, so search speed is than very fast.Because might be through the calculating of hash function, different keys is mapped on the same hash address, just produce conflict (the general way that manages conflict commonly used is to utilize the Hash bucket, and the key that the address is identical is put into a bucket, and then searches) in bucket.So for ashing technique, the design of hash function is very crucial, should select one to calculate simple and the relatively more uniform hash function of address distribution, reduce conflict as far as possible.Douglas C.Schmidt proposed a kind of perfect Hash function (Perfect Hash Function) generation method, and the Hash function that this method generates can guarantee that the hash value that obtains does not conflict, and that is to say that each hash value all is unique.But the present defectiveness of going back of this method, experiment show that the Hash table size that keyword adds up to generation in 300 o'clock is 1042, and 3 times of also being the keyword sum are more; When keyword rose to 685, the Hash table size can rise to 3833, was that 5 times of keyword sum are also more than.Hence one can see that, and the Hash table that generates with this method when data volume is big can be very big, and the space waste can be very serious.And when data volume was big, it was extremely time-consuming to generate a perfect Hash function.
Search tree comprises B tree, B+ tree, Trie tree etc. and their various distortion.Different tree roots is applied in the different data environments according to the characteristics of self.The data structure of index tree is comparatively speaking than other index structure complexity, and still good search tree method is used under the suitable environment and can has very high search efficiency.
The search efficiency of Trie tree is just than higher, and is relevant with the time and key self and the length thereof of a key of Trie tree search, is O (1) the soonest, and promptly the ground floor in tree can judge whether to search, and the worst situation is O (n), and n is the number of plies of Trie tree.Since many times the most of node branch of Trie tree seldom, so the waste of Trie tree construction space is many.And work as Trie tree node branches very for a long time, carry out binary search and can increase the more time expense again.
For a dictionary, most important function is searched exactly, in especially using as this class of natural language processing, needs frequent inquiring about to obtain word information in dictionary.So the efficient of data search is very important concerning dictionary methods.
Be used to organize the most frequently used indexing means of dictionary to mainly contain two kinds at present, a kind of is the Trie index tree, and a kind of is hash (Hash) list structure.
Based on the Dictionary Mechanism of Trie index tree generally is that lead-in to speech calculates hash value, sets up lead-in hash and shows, and sets up the Trie index tree then.Only need during retrieval word for word mate along the tree chain, need not predict the length of waiting to look into speech, it is only relevant with the length of this speech to search the required time of speech.But because the word branch is many, so the waste of the space of Trie index tree is more serious.Have experimental result to show, the dictionary of Trie tree construction take up room than the dictionary based on the Hash index take up room big on nearly one times.
Dictionary Mechanism based on the Hash index is exactly the hash value that a kind of Hash function of structure calculates word, and a bucket put in the word that hash value is identical, calculates the hash value of waiting to look into speech during retrieval earlier, enters then in the corresponding Hash bucket and carries out binary chop.Commonly used have lead-in Hash and two word Hash methods.Lead-in Hash utilizes lead-in to generate hash value, and the speech that hash value is identical is placed in the same Hash bucket, carries out binary chop then in bucket.Two word Hash then are a kind of improvement on lead-in Hash basis.Present two word Hash have two kinds of methods, and the one, the Hash index set up successively in turn in preceding two words of word, the formation degree of depth is 2 Trie subtree, remaining character string is entered in the corresponding Hash bucket again carry out binary chop; The 2nd, calculate hash value according to preceding two word ISNs, set up one-level Hash concordance list, enter then in the identical table of hash value and carry out binary chop.It is O (log that the dictionary of employing Hash index mechanism is searched time complexity
2N), N is that maximal value counted in the identical entry of hash value.
Summary of the invention
The present invention proposes a kind of dictionary management and search method based on the perfect even numbers group TRIE Tree structure.Convert the Trie tree construction to two linear array representations, and a kind of optimisation strategy has been proposed in construction process, when handling the Trie tree node, current active node is sorted by the branch node number, the node that priority processing branch node number is maximum is to reduce the sparse degree of array.Proposing a kind of adaptive coding scheme simultaneously, is that the coding unit carries out automatic coding to character with the byte, makes the present invention not only can support various coding schemes, and can further improve space availability ratio.At first with the Trie tree representation of the lexical item in the dictionary, convert corresponding array based on the dictionary of perfect even numbers group TRIE Tree structure then to.Retrieval to lexical item in this dictionary is carried out on the structure of array.At most only need carry out n-1 integer addition can obtain the result to the speech that to retrieve a length be n character.Retrieval time, complexity was O (n).
The objective of the invention is in order to solve the existing shortcoming of Trie tree construction, provide a kind of higher than existing dictionary methods search efficiency, still less dictionary management and search method take up room.
1. even numbers group TRIE Tree structure
The present invention is simplified to two linear arrays with the Trie tree construction, as shown in Figure 1.
Element in base array and the check array is one to one, each element in the base array is equivalent to the node of Trie tree, its value is done the base value of state transitions, and the value of relevant position is equivalent to proof test value in the check array, is used to check whether this state exists.For a transfer, must satisfy following two conditions from state s to state t:
1.base[s]+c＝t
2.check[t]＝s
Wherein c is an input variable.
Make that i is an array index, base[i] and check[i] be at 0 o'clock and represent this position for empty, base[i] but represent that this state is a done state during for negative value.The building method of two arrays is as follows:
Suppose state AB is arranged
1, AB
2, AB
3... AB
n, state A is designated as i under in array.For positive integer k, if
Base[k+B
1]=0, check[k+B
1]=0, base[k+B
2]=0, check[k+B
2]=0 ..., base[k+B
n]=0, check[k+B
n]=0, then the base value base[i of A]=k.
K that is to say, as long as can make the direct descendent of A can both put into array.Base[i] after value determines, state AB
1, AB
2, AB
3... AB
nSubscript in array is determined immediately, is respectively k+B
1, k+B
2, k+B
3..., k+B
n,
And check[k+B
1]=check[k+B
2]=...=check[k+B
n]=i.
The array structure will be searched a key Ab after finishing, and only needs to judge check[base[A]+b] whether equal A, if represent that then Ab searches in the Trie tree, otherwise failure is returned in search.
Even numbers group Trie tree (Double-Array Trie) method has effectively reduced the space waste of Trie tree construction, but still also can exist bigger data sparse in the array of utilizing this method to generate.In order further to reduce the waste of array space, the present invention optimizes the method for structure array, adds a kind of ordering strategy in the structure array, promptly all handles the maximum node of current branch node each time earlier.
2. the optimization access strategy of perfect even numbers group TRIE Tree
Consider that in even numbers group Trie tree method the position of each node in array all is by the base value decision of laststate just of its father's node.And its base value of node determine to depend on the current clear position of array and the direct descendent of this node.The direct descendent of a node is many more, and the conflict that this node is run into when looking for the base value is also just many more.Therefore the more node of priority processing branch helps reducing conflict, and it is excessive to avoid array to increase, and it is sparse to reduce data.
If a vocabulary is: " aa, aab, aad, bc, be, bed, cd ", then the Trie of this vocabulary correspondence tree is as Fig. 2.
Building method after the employing optimisation strategy is as follows:
1. read in lexicon file, structure Trie tree;
2. initialization active junction point tabulation adds this tabulation to the ground floor node.
3. if the active junction point tabulation is not empty, then choosing the maximum node of direct descendent number in the active junction point tabulation is current node.Otherwise method finishes, and the array structure is finished.
4. the visit current node determines its value in the base array, determines the position of each direct descendent in two arrays, makes that the value of each direct descendent in the check array is the array index of current node.
5. the direct descendent with current node adds the active junction point tabulation.Repeating step 2.
It is as follows to have constructed the corresponding search method in back:
1. make x equal the initial character sequence number.
2. read in input variable c, the sequence number of c is y.
3.t:＝base[x]+y
If?check[t]＝＝x?then
x:＝t
else?fail
endif
4. if base[t] not for negative, repeating step 2.Otherwise, but t is a done state.
Retrieving a speech only need carry out simple integer addition several times and get final product.The retrieval expense is only relevant with term length, and the method time complexity is O (n), and n is that the speech of term is long.
3. the self-adaptation character code of perfect even numbers group TRIE Tree is optimized
Can see, before with dictionary construction being even numbers group Trie tree, must determine the expression mode of character in even numbers group Trie tree construction.In TRIE tree and the even numbers group TRIE Tree transfering state be input as character, common way is directly to adopt the value of character to retrieve the location, such as directly adopting ASCII character, calculates its ISN for Chinese character, final input value as state transitions.Because current text code is of a great variety, the present invention proposes the character-coded optimisation strategy of a kind of self-adaptation, can support various language and coding scheme.Its basic ideas are: in order to reduce the side-play amount of high frequency character state transitions, the character code value that frequency of utilization is high more is more little.Concrete method is:
1. the character in the scanning vocabulary is added up word frequently;
2. sort toward low from high frequently according to word;
3. the character code value is the sequence number of its ordering.
If a vocabulary is: " aa, aab, aad, bc, be, bed, cd ", then word is shown as follows frequently:
| Character | Word frequently | Encoded radio |
| ?a | ?6 | ?1 |
| ?b | ?4 | ?2 |
| ?c | ?2 | ?4 |
| ?d | ?3 | ?3 |
| ?e | ?2 | ?5 |
Therefore each input variable value is the sequence number of input character.In order to adapt to double-byte characters coding (as GB2312, BIG5), we have introduced the encoded radio of the integer array Charset store character of a 64K.Computing formula is:
Code=Charset[Char], wherein Char is the round values of character correspondence, that is: the low level of the high-order * 256+ double-byte characters of the Byte value of single-byte character or double-byte characters.The encoded radio that Code comes out for storage.
This encoding scheme need not to revise the code of dictionary methods under the different situation of input text encoding scheme.And it is sparse further to reduce array, improves space availability ratio.
4. the performance test of perfect even numbers group TRIE Tree
For the Chinese dictionary of 80283 entries, we adopt the perfect even numbers group TRIE Tree method, and space availability ratio is 91.71%, per second on average can be retrieved 163.5 ten thousand of various entries at random, utilize this dictionary to carry out the maximum match participle, unit (CPU3.0G, internal memory 1G) speed is 23.81MB/s.
A kind of dictionary with perfect even numbers group TRIE Tree management and search method comprise step:
Step 1 is a coding unit to the entry in the dictionary with two bytes, and the frequency that occurs of the unit of coding respectively in the statistics dictionary gives different coding unit different sequence codes according to the frequency size; The coding unit big more to the frequency of occurrences in dictionary, the sequence code of giving is more little, at last all coding units and corresponding sequence code is kept in the sequence code file;
The array of base value described in the step 2 is corresponding one by one with element in the proof test value array; An element in the base value array is corresponding to the node of Trie tree, and this element value is as the base value of state transitions; The element value of correspondence position is used to check as proof test value whether this state exists in the proof test value array.
Described in the step 2 whole Trie tree is converted in the process of two integer arrays, current active node is sorted, all handle the maximum node of current branch node each time earlier.
The present invention uses the even numbers group TRIE Tree structure, dictionary is represented with the Trie tree construction, the Trie tree converts two linear arrays to, with the byte numerical approach and the means that unit encodes to the lexical item in the dictionary again, overcome existing dictionary method inquiry shortcoming, solve the problem of dictionary management and retrieval.Reach the raising search efficiency, reduce the effect in space.
Description of drawings
Fig. 1 is that the present invention is simplified to two linear number picture groups with the Trie tree construction.
Fig. 2 is the Trie tree synoptic diagram of vocabulary correspondence.
Fig. 3 is the process flow diagram of dictionary with perfect even numbers group TRIE Tree management of the present invention and search method.
Embodiment
Fig. 1 is the synoptic diagram that the Trie tree construction is converted into even numbers group Trie tree construction.
Fig. 2 represents all speech in the dictionary with the Trie tree construction.
Fig. 3 is the flow process of dictionary with perfect even numbers group TRIE Tree management of the present invention and search method,
Its step:
Step 1 is a coding unit to the entry in the dictionary with two bytes, and the frequency that occurs of the unit of coding respectively in the statistics dictionary gives different coding unit different sequence codes according to the frequency size; The coding unit big more to the frequency of occurrences in dictionary, the sequence code of giving is more little, at last all coding units and corresponding sequence code is kept in the sequence code file;
Claims (3)
1. a dictionary with perfect even numbers group TRIE Tree is managed and search method, comprises step:
Step 1 is a coding unit to the entry in the dictionary with two bytes, and the frequency that occurs of the unit of coding respectively in the statistics dictionary gives different coding unit different sequence codes according to the frequency size; The coding unit big more to the frequency of occurrences in dictionary, the sequence code of giving is more little, at last all coding units and corresponding sequence code is kept in the sequence code file;
Step 2, read in all entries of dictionary, use the Trie tree representation, read in the sequence code file,, handle all Trie tree nodes successively according to the strategy of the more node priority processing of branch node number, the most whole Trie tree converts two integer arrays to, one is the base value array, and one is the proof test value array, and two arrays are saved as file;
Step 3, first sequence code of encoding unit of term of submitting to the user is the base value array index, follow-up coding unit is successively as input variable, the base value of current state adds that input variable is as the subscript of next state in array, if the proof test value of current state equals the array index of laststate, then continue to judge NextState, finally return and retrieve successfully or fail.
2. the method for claim 1 is characterized in that: the array of base value described in the step 2 is corresponding one by one with element in the proof test value array; An element in the base value array is corresponding to the node of Trie tree, and this element value is as the base value of state transitions; The element value of correspondence position is used to check as proof test value whether this state exists in the proof test value array.
3. the method for claim 1 is characterized in that: described in the step 2 whole Trie tree is converted in the process of two integer arrays, current active node is sorted, all handle the maximum node of current branch node each time earlier.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101306903A CN100444167C (en) | 2005-12-21 | 2005-12-21 | Method for managing and searching dictionary with perfect even numbers group TRIE Tree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101306903A CN100444167C (en) | 2005-12-21 | 2005-12-21 | Method for managing and searching dictionary with perfect even numbers group TRIE Tree |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1786962A CN1786962A (en) | 2006-06-14 |
| CN100444167C true CN100444167C (en) | 2008-12-17 |
Family
ID=36784419
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005101306903A Active CN100444167C (en) | 2005-12-21 | 2005-12-21 | Method for managing and searching dictionary with perfect even numbers group TRIE Tree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100444167C (en) |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100495402C (en) * | 2006-12-31 | 2009-06-03 | 中国科学院计算技术研究所 | Method for constructing perfect Hash function for processing bulk dictionary |
| CN101398830B (en) * | 2007-09-27 | 2012-06-27 | 阿里巴巴集团控股有限公司 | Thesaurus fuzzy enquiry method and thesaurus fuzzy enquiry system |
| CN101916263B (en) * | 2010-07-27 | 2012-10-31 | 武汉大学 | Fuzzy keyword query method and system based on weighing edit distance |
| CN102521418A (en) * | 2011-12-31 | 2012-06-27 | 青岛海信宽带多媒体技术有限公司 | Pinyin storage structure and pinyin input method |
| CN102651026B (en) * | 2012-04-01 | 2015-02-18 | 百度在线网络技术（北京）有限公司 | Method for optimizing word segmentation of search engine through precomputation and word segmenting device of search engine |
| CN103823814B (en) * | 2012-11-19 | 2017-12-01 | 腾讯科技（深圳）有限公司 | A kind of information processing method and device |
| CN103365992B (en) * | 2013-07-03 | 2017-02-15 | 深圳市华傲数据技术有限公司 | Method for realizing dictionary search of Trie tree based on one-dimensional linear space |
| CN103365991B (en) * | 2013-07-03 | 2017-03-08 | 深圳市华傲数据技术有限公司 | A kind of dictionaries store management method realizing Trie tree based on one-dimensional linear space |
| CN104331397B (en) * | 2014-06-19 | 2017-07-07 | 深圳市迪威泰实业有限公司 | A kind of machine translation method and system |
| CN107305567A (en) * | 2016-04-21 | 2017-10-31 | 北京智能管家科技有限公司 | A kind of sentence querying method and device based on double-deck trie trees |
| CN106250526A (en) * | 2016-08-05 | 2016-12-21 | 浪潮电子信息产业股份有限公司 | A kind of text class based on content and user behavior recommends method and apparatus |
| CN108021569A (en) * | 2016-11-01 | 2018-05-11 | 中国移动通信有限公司研究院 | The structure of AC automatic machines and Chinese multi-model matching method and relevant apparatus |
| CN108628907B (en) * | 2017-03-24 | 2021-09-17 | 北京京东尚科信息技术有限公司 | Method for matching Trie tree with multiple keywords based on Aho-Corasick |
| CN109426358B (en) * | 2017-09-01 | 2023-04-07 | 百度在线网络技术（北京）有限公司 | Information input method and device |
| CN108509505B (en) * | 2018-03-05 | 2022-04-12 | 昆明理工大学 | Character string retrieval method and device based on partition double-array Trie |
| AU2019203850B2 (en) * | 2019-03-04 | 2021-09-16 | Advanced New Technologies Co., Ltd. | Constructing blockchain world state merkle patricia trie subtree |
| CN110347804B (en) * | 2019-07-22 | 2023-05-12 | 同方知网数字出版技术股份有限公司 | Sensitive information detection method of linear time complexity |
| CN110888886B (en) * | 2019-11-29 | 2022-11-11 | 华中科技大学 | Index structure, construction method, key value storage system and request processing method |
| CN111309970A (en) * | 2020-03-27 | 2020-06-19 | 北京奇艺世纪科技有限公司 | Data retrieval method and device, electronic equipment and storage medium |
-
2005
- 2005-12-21 CN CNB2005101306903A patent/CN100444167C/en active Active
Non-Patent Citations (4)
| Title |
|---|
| 汉语词典快速查询算法研究. 李江波，周强，陈祖舜.第二届全国学生计算语言学研讨会论文集，第2004年卷. 2004 |
| 汉语词典快速查询算法研究. 李江波，周强，陈祖舜.第二届全国学生计算语言学研讨会论文集，第2004年卷. 2004 * |
| 用于汉字拼音输入的类trie树及其改进. 王博文，苏国辉.小型微型计算机系统，第2002年卷第6期. 2002 |
| 用于汉字拼音输入的类trie树及其改进. 王博文，苏国辉.小型微型计算机系统，第2002年卷第6期. 2002 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1786962A (en) | 2006-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100444167C (en) | Method for managing and searching dictionary with perfect even numbers group TRIE Tree | |
| US8572126B2 (en) | Systems and methods for optimizing very large n-gram collections for speed and memory | |
| US7031910B2 (en) | Method and system for encoding and accessing linguistic frequency data | |
| Shang et al. | Tries for approximate string matching | |
| US6816856B2 (en) | System for and method of data compression in a valueless digital tree representing a bitset | |
| US6785687B2 (en) | System for and method of efficient, expandable storage and retrieval of small datasets | |
| CN107153647B (en) | Method, apparatus, system and computer program product for data compression | |
| CN1552032B (en) | Database | |
| JP5858432B2 (en) | Method, system, and computer program product for providing a distributed associative memory base | |
| CN102142038B (en) | Multi-stage query processing system and method for use with tokenspace repository | |
| Itoh et al. | An efficient method for in memory construction of suffix arrays | |
| EP1234258B1 (en) | System for managing rdbm fragmentations | |
| US20130141259A1 (en) | Method and system for data compression | |
| CN102411580B (en) | The search method of XML document and device | |
| CN108197313B (en) | Dictionary indexing method for realizing space optimization through 16-bit Trie tree | |
| CN101398830B (en) | Thesaurus fuzzy enquiry method and thesaurus fuzzy enquiry system | |
| Guthrie et al. | Storing the web in memory: Space efficient language models with constant time retrieval | |
| CN101576929A (en) | Fast vocabulary entry prompting realization method | |
| Conway et al. | Optimal hashing in external memory | |
| Pibiri et al. | Efficient data structures for massive n-gram datasets | |
| CN105335481A (en) | Large scale character string text suffix index building method and device | |
| Jansson et al. | Linked dynamic tries with applications to LZ-compression in sublinear time and space | |
| Flor | A fast and flexible architecture for very large word n-gram datasets | |
| CN101944086A (en) | Whole word index dictionary | |
| US7672925B2 (en) | Accelerating queries using temporary enumeration representation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| ASS | Succession or assignment of patent right |
Owner name: BEIJING SKYVEIN NET COMPUTER CO., LTD. Free format text: FORMER OWNER: INSTITUTE OF COMPUTING TECHNOLOGY, CHINESE ACADEMY OF SCIENCE Effective date: 20120406 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 100080 HAIDIAN, BEIJING TO: 100190 HAIDIAN, BEIJING |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20120406 Address after: 100190, room 225, North building, No. 6 science South Road, Haidian District, Beijing, Zhongguancun Patentee after: Beijing Skyvein Net Computer Co., Ltd. Address before: 100080 Haidian District, Zhongguancun Academy of Sciences, South Road, No. 6, No. Patentee before: Institute of Computing Technology, Chinese Academy of Sciences |
|
| ASS | Succession or assignment of patent right |
Owner name: LING-JOIN ZHONGKE SOFTWARE (BEIJING) CO., LTD. Free format text: FORMER OWNER: BEIJING SKYVEIN NET COMPUTER CO., LTD. Effective date: 20131218 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 100190 HAIDIAN, BEIJING TO: 100081 HAIDIAN, BEIJING |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20131218 Address after: 100081 Beijing Haidian District Suzhou Street No. 49-3 hospital Ying Zhi Building 5 Patentee after: Ling Zhong Zhongke software (Beijing) Co., Ltd. Address before: 100190, room 225, North building, No. 6 science South Road, Haidian District, Beijing, Zhongguancun Patentee before: Beijing Skyvein Net Computer Co., Ltd. |