CN103577555A - Big data analysis method based on internet of vehicles - Google Patents
Big data analysis method based on internet of vehicles Download PDFInfo
- Publication number
- CN103577555A CN103577555A CN201310494123.0A CN201310494123A CN103577555A CN 103577555 A CN103577555 A CN 103577555A CN 201310494123 A CN201310494123 A CN 201310494123A CN 103577555 A CN103577555 A CN 103577555A
- Authority
- CN
- China
- Prior art keywords
- attribute
- node
- split
- attribute list
- gini
- 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
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a big data analysis method based on internet of vehicles. The big data analysis method comprises the following steps: completely splitting input data records to be in the form of attribute lists, sorting for one time, splitting root nodes, completely splitting the attribute lists through circulating splitting, generating a complete decision tree, extracting a tree structure, and generating corresponding mode rules. According to the big data analysis method, data of masses of vehicles can be analyzed; when a request searched by a user is received, data and data combinations required by a user can be quickly analyzed, so that decision aids are provided for the user.
Description
Technical field
The present invention relates to a kind of data analysis searching method, relate in particular to a kind of large data analysing method based on car networking.
Background technology
Along with continuing to increase of automobile pollution, consumer is to automotive safety increase in demand, car owner is not only the safe reliability of vehicle itself to the demand of safety, also being embodied in those can provide by electronic security(ELSEC) product or service technology the DAS (Driver Assistant System) of more safety guarantee, the expansion of this demand to drive market to expand gradually to automotive safety detection demand.
The car networking product of currently available technology are as Anji star, sync etc., mainly on concentrated car-mounted terminal, there is limitation in the service providing for car owner, each is served link and does not network or network and fail organically to integrate, thereby cause series of problems: when vehicle failure, car owner can only rely on rescue, the maintenance in 4S shop; Due to the loss of vehicle failure field data, increased 4S shop maintenance cost.Insurance company, because there is no accurate data, is happened occasionally by insurance fraud phenomenon, loses huge; And the car owner who has good driving habits does not obtain the excellent expense of objective premium etc.
Per second can the generation of each car that forms car networking reaches 100 data, number will form magnanimity large database concept year in year out in automobile necessarily, car networking is combined with large data, car networking data can be formed to huge " data treasure-house ", can be based on this, develop the commercial applications of Huge value, and do not have in the market the car large intelligent data analysis Related product of networking.
Summary of the invention
Embodiment of the present invention technical matters to be solved is, a kind of large data analysing method based on car networking is provided.Can be to the large data analysis of vehicle and fast search.
In order to solve the problems of the technologies described above, the embodiment of the present invention provides a kind of large data analysing method based on car networking, comprises the following steps:
The status data collection about vehicle collecting is split into attribute list, by SPRINT algorithm, Continuous valued attributes table is sorted, and carry out and calculate Gini value task and find split point from described attribute list, the attribute list with same alike result name be attached on corresponding node and stamp the mark of respective nodes, and being distributed to same Reduce and processing;
The attribute list that includes corresponding Gini value and split point is compared to the size of Gini value by Reduce, use the attribute of minimum Gini value as best Split Attribute, the Id that records that corresponding split point is split into the attribute list of same node writes Hash table, again attribute list is exported, and root node is inputted in HDFS file;
By Reduce, according to the feature of attribute, construct different histograms, and by the scanning to ordering Continuous valued attributes table or categorical attribute table, real-time renewal histogram, calculate corresponding division Gini index, thereby find the best splitting point of the current attribute of present node, the information of output is all put into HDFS file;
Attribute list on different nodes is distributed, all properties table of present node is processed by Reduce, and whether identify present node be leaf node, nonleaf node is carried out to circulating filtration, and present node is write in HDFS file as leaf node information.
Further, the OBD data that the status data collection of described vehicle obtains vehicle by obtaining car-mounted terminal generate.
Further, described OBD data also comprise tire pressure, locator data.
Further, described attribute list comprises following form: <<0, Ai, AC, Vj>, <idj, Cj>>; Wherein 0 represents that all attribute lists are attached on root node, Ai represents the attribute of the i of training set, AC represents the classification of attribute, and idj represents that every attribute list is recorded in the recording indexes of whole data centralization, and Cj represents the value of the generic attribute of this attribute list record.
Further, described in to be distributed to that same Reduce processes be to define Patitioner, and the middle key-value pair by having different Ai that will define key is distributed to different reducer.
Further, described in, be distributed to same Reduce to process run-out key value representation be <<0, Ai, AC, Vj>, <idj, Cj, Gini, Split>>; Wherein Gini represents the Gini value of respective attributes table, and Split represents split point.
Further, whether described identification present node is that leaf node is identified as leaf node by following steps:
The all properties table of present node belongs to same class;
The contained attribute list quantity of present node is less than predefined threshold values;
When all attribute lists are all split into same node;
The attribute list of present node splits in two child nodes.
Implement the embodiment of the present invention, there is following beneficial effect: by the present invention, can, from the data of the vehicle of magnanimity, carry out analyzing and processing, when receiving user's searching request, can express-analysis go out the data of client's needs and the combination of data, thereby provide decision-making auxiliary to user.
Embodiment
For making the object, technical solutions and advantages of the present invention clearer, below the present invention is described in further detail.
The present invention is according in actual should using, the features such as data volume is huge, employing, based on Hadoop(distributed system platform, has extremely strong distributed storage and the ability of calculating) MIIS(Multi-level Inverted Index Structure) inverted index structure, AMPS(Align and Merge Placement Strategy) inverted index and Replica Placement Strategy and PforDelta boil down to basic system platform.
MIIS inverted index structure, basic thought is that the main contents piecemeal of inverted index file is left in back end, in title node, deposit the mapping of keyword and piece positional information, and sometime section collection of document and the mapping of its inverted index place data block location information, inverted index piece is stored on each back end in cluster and is safeguarding the status information of the data block set of depositing inverted index.
The basic thought of the distributed structure algorithm of inverted index is: first, to collection of document, divide block operations, the piece having divided is left in HDFS file system; Then utilize Map operation to generate inverted index file to each document piece, utilize Reduce operation to merge all inverted index files, generate the inverted index of the document collection; Finally, inverted index is written in HDFS file system by piecemeal according to AMPS Placement Strategy, upgrade the auxiliary block information index on back end simultaneously, and send more nonproductive poll index and the auxiliary index of deleting on newname node of corresponding data to title node.Data in whole building process read and deposit is all to move in HDFS file system,
Utilizing MapReduce can be the distributed structure inverted index of large volume document piece rapidly, and place inverted index file according to AMPS Placement Strategy, simultaneously with MIIS inverted index structure, organize inverted index file, for inquiry and batch updating operate, lay a good foundation.
Hadoop mainly contains HDFS file system and two major parts of MapReduce programming model form.
HDFS (Hadoop distributed file system) file system framework is the structure that adopts supvr-worker (Master-Slave), namely a title node (Master) and a plurality of back end (Slave).
MapReduce is also the pattern that adopts supvr-worker (Master-Slave).Hadoop is divided into a lot of little tasks (task) an operation (job) to be carried out, and this is comprising two kinds of tasks: Map task and Reduce task.
The present invention adopts SPRINT algorithm that the OBD data set of the car-mounted terminal collecting is processed, and minute three phases execution analysis is processed, and wherein the first stage is mainly used to produce ordering attribute list, produces the division of root node.Subordinate phase is mainly used to carry out the fission process of circulation, completes the division completely of attribute list, generates a complete decision tree.Phase III is mainly used to extract tree construction, and generates corresponding pattern rules.
First stage: in this stage, the data recording of input need to be split into completely to the form of attribute list, and once sort.In this process, be ready to attached all properties lists on Node0, carry out corresponding calculating, calculate Gini value, the split point of every attribute list, then comprehensively compare, realize division first.Then I am attached to corresponding attribute list on Node00 and Node01.For the circulation of subordinate phase ready.First stage need to be carried out two Job and be completed.
The map process of Job1:
This process splits OBD data set, and generates attribute list completely, is used for characterizing all data sets.Training set is split as to M part, for every a training set, there is identical attribute number n.This process is by InputFormat(input format) to realize, it is responsible for resolving and generating the input key-value pair of map ().For map () method, be input as every record of data set, by defining corresponding rule, output is expressed as: <<0, Ai, AC, Vj>, <idj, Cj>>.0 represents that all attribute lists are attached on root node, Ai represents the attribute of the i of training set, AC represents the classification (classification or successive value) of attribute, and idj represents that every attribute list is recorded in the recording indexes of whole data centralization, and Cj represents the value of the generic attribute of this attribute list record.The <<0 with same alike result name, Ai, AC, Vj>, <idj, Cj>> is distributed to identical reduce, and according to the compareTo () of key definition, carries out sequence before arriving reduce.It is called as middle key-value pair.The quantity that middle key assignments pair set is divided into and the quantity of reduce equates, here namely equates with the quantity of attribute.Identical attribute can be mapped in corresponding reducer and process.In this computing, definition Patitioner, and the middle key-value pair by having different Ai that will define key is distributed to different reducer.
The Reduce process of Job1:
When Reduce receives all attribute lists, MapReduce has carried out sequence.The main task of Reduce is:
1. calculate the best splitting point of every attribute list, corresponding Gini value.To utilize the best Split Attribute of these acquisition of informations.
2. the results such as the Gini value of attribute list, split point are attached to the enterprising line output of attribute list
In this Reduce process, for the information such as Gini value and split point are attached on attribute list.The output key-value pair of this process reduce can be expressed as (<<0, Ai, AC, Vj>, <idj, Cj, Gini, Split>>).
Gini represents the Gini value of respective attributes table, and split represents split point.Obvious, because the attribute list of exporting is to be all attached to node Node0 above, directly use 0 mark present node here.And after circulation in, by 0 label that changes current tree node into.
The map process of Job2:
The output that is input as Job1 in this stage, includes all attribute lists of corresponding Gini value and split point.The map here does not need to do special operation, owing to only there is now a node, i.e. and root node, thus only all attribute lists need to be distributed in a reduce, to carry out fission process.Here, the Reduce number of Job is made as to 1, only with a Reduce, achieves the goal.
The Reduce process of Job2:
The Reduce of this process will be mainly used to divide.For the output of map above, i.e. the input of this Reduce, all includes corresponding Gini value and split point.Reduce, by comparing the size of Gini value, can be used the attribute with minimum Gini value as Split Attribute.After determining best Split Attribute, by introducing this data structure of Hash table.After determining best Split Attribute, for this Split Attribute, adopt corresponding split point, the Id that records that splits into the attribute list of same node is write to Hash table.Like this, use Hash table, for the attribute list of non-Split Attribute, by sweep record Id, whether be included in Hash table, determine to be split in Node00 or Node01.In fact, by this Reduce process, obtained being attached to all properties list on Node00 and Node01---in input HDFS file.During output, except output attribute list, we also will export the character of root node, that is: nonleaf node, Split Attribute, split point.These values can complete mark decision tree root node., use the multifile output function of Reduce here, the information of root node is write in the file of appointment.Like this, when division finishes, could find the tree generating.
Subordinate phase: the first stage has been prepared all properties table of Node01 and Node00, has exported the information of node 0 in the file of the expression tree of appointment.In subordinate phase, need to define the iteration of a node split so, until all node splits complete.Same, each process of iteration, all needs 2 Job to complete.It should be noted that the input as job4 in the output of subordinate phase Job3, and the output of job4 is as the input of job3, produces thus circulation.When all attribute lists have divided, while being in fact assigned to corresponding leaf node, circulation stops, and enters the phase III.
The map process of Job3:
The function of Map is mainly used to realize the mapping of attribute list.
Map is reading out data from the file of HDFS always.So, Map must be used the structure of the attribute list of definition before the Information generation that reads, to utilize and the identical Partitioner of Patitioner of first stage use realizes the distribution of attribute list of the different attribute name of different nodes.This stage adopts paralleling tactic completely.In addition, Map does not do other processing.
Reduce process in Job3:
After arriving reducer by key-value pair in the middle of obtaining after using the distribution rules of the Map of Job3 to process, a Redeuce receives the attribute list with same alike result name of same node.Same, Continuous valued attributes has also passed through sequence.So, just as the reducer of first stage Job1, the reducer here can belong to numeric type or classifying type according to the feature of attribute, constructs different histograms.
By the scanning to ordering Continuous valued attributes table or categorical attribute table, real-time renewal histogram, calculates corresponding division Gini index, thereby finds the best splitting point of the current attribute of present node.Because the information of output is all put into file, the available any simple form of output of this stage Reduce, but need to comprise following information (<0, Ai, AC, Vj, idj, Cj, Gini, Split>) completely.The existing identical feature of output of this Reduce in first stage Job2.
The map process of Job4:
In Job4, Map receives after the output of Reduce in Job3 from HDFS file, same execution distribution.But, different from Map in Job3 here: only the attribute list being attached on different nodes to be distributed, and do not considered whether they have identical attribute-name.In order to reach such object, also can improve the efficiency of algorithm simultaneously, reduce the complexity of data structure, only need the present node of the attribute list of simple input to be set to key, and other all information are put into value.Like this, the attribute list of all different attributes in same node (being accompanied with the information such as division Gini value, split point of corresponding attribute) will be distributed to same Reducer.The form shfft of the middle key-value pair of this Map output is shown: (<<Node>, <Ai, AC, Vj, idj, Cj, Gini, Split>>).The Node here represents present node.
The reduce process of Job4:
This process, except being used for carrying out the division of the attribute list that present node is comprised, is also used for attribute list to filter.When all properties table of present node enters Reduce, and while all having comprised the best splitting point of different attribute and Gini value.First should judge: whether present node is leaf node.If present node is to belong to leaf node, all properties table of present node does not just divide so, does not enter circulation next time, the filtration that Here it is to attribute list.Meanwhile, the information using present node as leaf node write in the file of expression tree of appointment.
When a node is considered to leaf node, so necessarily exist following situation:
1. all properties table of present node belongs to same class.
2. the contained attribute list quantity of present node is less than our predefined threshold values.
3. when all attribute lists are all split into same node.This situation is due to split point, to be arranged in the centre of the attribute list record with the value that Continuous valued attributes is identical, and this value is maximal value or minimum value in attribute.
4. the attribute list of present node splits in two child nodes.But the attribute list number being attached in these two child nodes is all less than our predefined threshold values.And, the maximum quasi-equal comprising in all properties table in these two child nodes.
If a kind of in four kinds of situations above appears in present node, present node is considered to leaf node immediately so.If any one in above-mentioned situation do not appear in present node, present node is thought nonleaf node immediately so.
If present node is identified nonleaf node, division will be proceeded.The division Gini index of same relatively each attribute, finds best Split Attribute, and split point, is attached to attribute list in corresponding child node, writes in the file that represents tree using the information of present node as nonleaf node.
The termination of circulation:
When the division of all attribute lists completes, i.e. the construction complete of tree.At this moment, program there will be two features:
1) all attribute lists are filtered, and do not have any attribute list and enter in Job3.
2) in the layer of present tree, all nodes are all leaf nodes, there is no nonleaf node.
So, can utilize any one end loop in above-mentioned two situations.In the present embodiment, we adopt second method, judge whether to exist nonleaf node to carry out end loop.
Phase III: when subordinate phase completes, all attribute lists have all been assigned in corresponding leaf node, and at this time, the structure of tree completes.Because the output of MapReduce is directly written in the file of distributed system, so the task of phase III is to obtain the structure of tree from these reduce files.In the Job2 of first stage and the Job4 of subordinate phase, the information of output present node is in the file of preassigned expression tree in HDFS.In whole process, be to show a tree by designing some special output informations.
1) when passing through judgement, find that present node is leaf node, to the form writing in file, be so: <<Node, vertex ticks >, <isleaf, maximum class >> on leaf node
2) when passing through judgement, discovery counts out money node while being nonleaf node, we to the form writing in file are: <<Node, vertex ticks >, <noleaf, Split Attribute, split point >>
Obviously, use the information of these all nodes, can from file, extract easily tree construction, because all information about tree has all obtained.All nodes of tree, comprise that leaf node and nonleaf node are all used " Node " mark.Nonleaf node is used " noleaf ", and leaf node is used " isleaf ".On nonleaf node, comprised the information of Split Attribute and split point, and on leaf node, comprised the value of maximum kind attribute, like this, one tree has been constructed.
When response user send searching keyword request time, on title node, by nonproductive poll index, find the piece list information of this keyword, then JobTracker sets up an inquiry job, to TaskTracker, initiate a plurality of Map tasks and a Reduce task, by map task, index block is inquired about, import Query Result into reduce task node and merge, finally return to the final Query Result of user.
Above disclosed is only a kind of preferred embodiment of the present invention, certainly can not limit with this interest field of the present invention, and the equivalent variations of therefore doing according to the claims in the present invention, still belongs to the scope that the present invention is contained.
Claims (7)
1. the large data analysing method based on car networking, is characterized in that, comprises the following steps:
The status data collection about vehicle collecting is split into attribute list, by SPRINT algorithm, Continuous valued attributes table is sorted, and carry out and calculate Gini value task and find split point from described attribute list, the attribute list with same alike result name be attached on corresponding node and stamp the mark of respective nodes, and being distributed to same Reduce and processing;
The attribute list that includes corresponding Gini value and split point is compared to the size of Gini value by Reduce, use the attribute of minimum Gini value as best Split Attribute, the Id that records that corresponding split point is split into the attribute list of same node writes Hash table, again attribute list is exported, and root node is inputted in HDFS file;
By Reduce, according to the feature of attribute, construct different histograms, and by the scanning to ordering Continuous valued attributes table or categorical attribute table, real-time renewal histogram, calculate corresponding division Gini index, thereby find the best splitting point of the current attribute of present node, the information of output is all put into HDFS file;
Attribute list on different nodes is distributed, all properties table of present node is processed by Reduce, and whether identify present node be leaf node, nonleaf node is carried out to circulating filtration, and present node is write in HDFS file as leaf node information.
2. the large data analysing method based on car networking according to claim 1, is characterized in that, the status data collection of described vehicle obtains vehicle OBD data by obtaining car-mounted terminal generate.
3. the large data analysing method based on car networking according to claim 2, is characterized in that, described OBD data also comprise tire pressure, locator data.
4. the large data analysing method based on car networking according to claim 1, it is characterized in that, described attribute list comprises following form: <<0, Ai, AC, Vj>, <idj, Cj>>; Wherein 0 represents that all attribute lists are attached on root node, Ai represents the attribute of the i of training set, AC represents the classification of attribute, and idj represents that every attribute list is recorded in the recording indexes of whole data centralization, and Cj represents the value of the generic attribute of this attribute list record.
5. the large data analysing method based on car networking according to claim 4, it is characterized in that, described to be distributed to that same Reduce processes be to define Patitioner, and the middle key-value pair by having different Ai that will define key is distributed to different reducer.
6. the large data analysing method based on car networking according to claim 4, it is characterized in that, describedly be distributed to same Reduce to process run-out key value representation be <<0, Ai, AC, Vj>, <idj, Cj, Gini, Split>>; Wherein Gini represents the Gini value of respective attributes table, and Split represents split point.
7. the large data analysing method based on car networking according to claim 1, is characterized in that, whether described identification present node is that leaf node is identified as leaf node by following steps:
The all properties table of present node belongs to same class;
The contained attribute list quantity of present node is less than predefined threshold values;
When all attribute lists are all split into same node;
The attribute list of present node splits in two child nodes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310494123.0A CN103577555A (en) | 2013-10-21 | 2013-10-21 | Big data analysis method based on internet of vehicles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310494123.0A CN103577555A (en) | 2013-10-21 | 2013-10-21 | Big data analysis method based on internet of vehicles |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103577555A true CN103577555A (en) | 2014-02-12 |
Family
ID=50049331
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310494123.0A Pending CN103577555A (en) | 2013-10-21 | 2013-10-21 | Big data analysis method based on internet of vehicles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103577555A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015062540A1 (en) * | 2013-10-31 | 2015-05-07 | 中国移动通信集团公司 | Driving amount model event-based storage and index methods and system |
| CN105373426A (en) * | 2015-07-28 | 2016-03-02 | 哈尔滨工程大学 | Method for memory ware real-time job scheduling of car networking based on Hadoop |
| CN108932255A (en) * | 2017-05-25 | 2018-12-04 | 北京万集科技股份有限公司 | A kind of vehicle integration capability analysis method and device |
| CN111695588A (en) * | 2020-04-14 | 2020-09-22 | 北京迅达云成科技有限公司 | Distributed decision tree learning system based on cloud computing |
-
2013
- 2013-10-21 CN CN201310494123.0A patent/CN103577555A/en active Pending
Non-Patent Citations (2)
| Title |
|---|
| 何元: "基于云计算的海量数据挖掘分类算法研究", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
| 杨宸铸: "基于Hadoop的数据挖掘研究", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015062540A1 (en) * | 2013-10-31 | 2015-05-07 | 中国移动通信集团公司 | Driving amount model event-based storage and index methods and system |
| CN105373426A (en) * | 2015-07-28 | 2016-03-02 | 哈尔滨工程大学 | Method for memory ware real-time job scheduling of car networking based on Hadoop |
| CN105373426B (en) * | 2015-07-28 | 2019-01-15 | 哈尔滨工程大学 | A kind of car networking memory aware real time job dispatching method based on Hadoop |
| CN108932255A (en) * | 2017-05-25 | 2018-12-04 | 北京万集科技股份有限公司 | A kind of vehicle integration capability analysis method and device |
| CN108932255B (en) * | 2017-05-25 | 2022-01-14 | 北京万集科技股份有限公司 | Vehicle comprehensive capacity analysis method and device |
| CN111695588A (en) * | 2020-04-14 | 2020-09-22 | 北京迅达云成科技有限公司 | Distributed decision tree learning system based on cloud computing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | Prediction of bus travel time using random forests based on near neighbors | |
| CN107577688B (en) | Original article influence analysis system based on media information acquisition | |
| CN105956015A (en) | Service platform integration method based on big data | |
| CN106484813B (en) | A kind of big data analysis system and method | |
| CN113590698B (en) | Artificial intelligence technology-based data asset classification modeling and hierarchical protection method | |
| CN102521364B (en) | Method for inquiring shortest path between two points on map | |
| CN109815254B (en) | Cross-region task scheduling method and system based on big data | |
| CN104679827A (en) | Big data-based public information association method and mining engine | |
| CN104133868B (en) | A kind of strategy integrated for the classification of vertical reptile data | |
| DE102012221251A1 (en) | Semantic and contextual search of knowledge stores | |
| CN112231333A (en) | Ecological environment data sharing and exchanging method and system | |
| CN103577555A (en) | Big data analysis method based on internet of vehicles | |
| CN104092744A (en) | Web service discovery method based on memorization service cluster mapping catalogue | |
| CN108763573A (en) | A kind of OLAP engines method for routing and system based on machine learning | |
| CN104268247A (en) | Master data imputation method based on fuzzy analytic hierarchy process | |
| CN103226609A (en) | Searching method for WEB focus searching system | |
| CN112966162A (en) | Scientific and technological resource integration method and device based on data warehouse and middleware | |
| CN113486983A (en) | Big data office information analysis method and system for anti-fraud processing | |
| Ma et al. | Driving style estimation by fusing multiple driving behaviors: a case study of freeway in China | |
| CN114064707A (en) | Data query method and device for data virtualization server and storage medium | |
| CN103455491A (en) | Method and device for classifying search terms | |
| CN103425748A (en) | Method and device for mining document resource recommended words | |
| CN113254517A (en) | Service providing method based on internet big data | |
| CN112967759B (en) | DNA material evidence identification STR typing comparison method based on memory stack technology | |
| CN116013084A (en) | Traffic management and control scene determining method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140212 |
|
| RJ01 | Rejection of invention patent application after publication |