2011 International Conference on Emerging Intelligent Data and Web Technologies

PoLyInfo: Polymer Database for polymeric materials design

Shingo Otsuka∗ , Isao Kuwajima† , Junko Hosoya† , Yibin Xu† and Masayoshi Yamazaki†
∗ Department of Information and Computer Sciences,
Kanagawa Institute of Technology, Kanagawa, Japan
Email: otsuka@ic.kanagawa-it.ac.jp
† Materials Database Station,
National Institute for Materials Science (NIMS), Ibaraki, Japan
Email: kuwajima.isao, hosoya.junko, xu.yibin, yamazaki.masayoshi}@nims.go.jp

Abstract—Polymer database (PoLyInfo) provides various As results of grappling with these problems, ‘PoLyInfo’
data required for polymeric materials design. The main data that is the polymer database system has been opened to the
source is academic papers on polymers. PoLyInfo collects public since April, 20013 . This system has the data model
information on polymer names, chemical structures, processing
methods of samples, measurement conditions, properties, used of the polymer and the function of the compound registry
monomers and polymerization methods. PoLyInfo continues system for the polymer. It goes continuing enhancing data
the expansion of data as one of NIMS materials database. and the system now.
This paper presents system architecture of ”Polymer database It is the first attempt to discuss cyclopaedically the sys-
(PoLyInfo)” and its distinctiveness. And we also report usage tem architecture, the search strategy and the usage survey
survey of our system.
in PoLyInfo. Though about nine years have passed since
Keywords-materials database; polymer database; chemical PoLyInfo was open to the public and there were some
structure search; polymer search reports concerning development concept and system survey
in PoLyInfo[5], [4], [6], [8], [10], [3].
I. I NTRODUCTION There is some large-scale database that relates to the poly-
We have opened NIMS materials database (MatNavi)1 in mer. For example, National Institute of Advanced Industrial
the Internet since April, 20032 as materials researchers and Science and Technology (AIST) provides spectrum database
apparatus designers obtain the information about a materials. of organic compound although there is not a lot of data
MatNavi is comprised of following things and we place it concerning the polymer. And Korea Research Institute of
with the portal site about materials. Chemical Technology (KRICT) provide ‘Chemical materials
Information Bank’ service. However we obtain the data of
– Data of basic properties about materials as atom, the materials made business in this system only because
molecules, crystal structure. the information is extracted from catalog data of the South
– Data of foundational engineering. Korea enterprise. There is materials information service of
– Search system and applications using database. ”MatWeb” in the United States. This service provides a lot of
We show a webpage of MatNavi in Figure 1 and main data such as ceramics and the metals besides the polymer.
databases in Table I. In this paper, we describe system However it is a data only of the materials made business
architecture, search method and usage survey of polymer offer because it uses the catalog data of the enterprise as
database (PoLyInfo) which has many entry data and users well as the example of KRICT. Besides this, there is service
in the MatNavi. of ”Polymers: A Property Database” that the publisher of the
There are various problems for the construction of the scientific journal offers. This service has aimed to make the
polymer database in consideration of materials design sup- user inspect the papers (pay contents). And the use fee of
port. First, the contents of entry data come in a variety of ”SciFinder” that is the article retrieval service of American
types. Moreover, it is necessary to design the database that Chemical Society is high. Hence the users can not use the
can systematically collect polymer information because the system airily.
polymer is composed of a lot of molecules, and there is In this way, these services provide useful information con-
a characteristic with complex relativity between molecules. cerning the materials but there are some problems (amount
For example, there are a view of materials and a view of of data is low, use fee is high, etc.). The advantage of our
compound of ethylene (monomer) in polyethylene that is system (PoLyInfo) is free access to a lot of polymer data
typical polymer. Therefore, it is necessary to store informa- extracted from articles.
tion of them in the databases. The structure of the compound
registry system (polymer dictionary) for the polymer is nec- II. S YSTEM ARCHITECTURE OF POLYMER DATABASE
essary to store polymer as compound in databases accurately. (P O LY I NFO )
Hence, it is possible to identify chemical structure and to PoLyInfo is composed of the simulation part (properties
accommodate fluctuation of description of trivial name. estimation) and the database part as shown in Figure 2.
1 http://mits.nims.go.jp/index en.html 3 NIMS manages this system now though Japan Science and Technology
2 NIMS is National Institute for Materials Science. Agency (JST) manages in March, 2003.

978-0-7695-4456-4/11 $26.00 © 2011 IEEE 22

DOI 10.1109/EIDWT.2011.13
 parameter are computable from the analytical result of the
real data.
In the database part, we divide polymer information into
properties data, structure information and polymerization
information and polymerization information cooperates with
the monomer database as the users do not have to con-
sider data model though it is internally maintained. About
information of name and structural unit, it is possible to
treat polymer which is different structure with a same
composition type using polymer dictionary. We call polymer
registered to PoLyInfo polymer sample and we employ
hierarchical data structure to express one polymer sample
efficiently.
Figure 1. Webpage of ”NIMS Materials Database”. The information about each polymer sample as the pri-
mary structure information that is registered in polymer
Table I dictionary and forming method is correspondence to every
NIMS M ATERIALS DATABASES (M AT NAVI ). polymer sample. And detail information about the real poly-
merization is included in the polymerization information.
No. Database Name
1 Database for Electronic Structures III. A STORAGE METHOD OF POLYMER DATA
2 Basic Database for Crystal Structures [Pauling File] In PoLyInfo, at first, we analyzed the polymer data using
3 3D Demo System for Ternary Phase Diagrams OMT method(OMT:Object Modeling Technique)[7], built
4 Diffusion Database
5 Superconducting Materials Database the data model and stored ObjectStore for build the polymer
6 Polymer Database [PoLyInfo] database in consideration of materials design support. After
7 NIMS Structural Materials Datasheet Onlines that time, we rebuilt this system using PostgreSQL because
(Creep, Fatigue, Corrosion, Space Use Materials Strength) object-relational database was diffuse.
8 Microstructures of Crept Materials
9 CCT Diagrams for Welding A. Analysis of the polymer information
10 Database System for Pressure Vessel Materials
11 Nuclear Materials Database Representative polymer information includes polymer
12 Nuclear Reaction Database for Materials name4 , Polymer classification5 , polymer properties6 . More-
Thermophysical Property Prediction System for Composites over, Solid state properties of polymer differ from stereo-
13 [Compo Therm] regularity, polymerization degree, forming method and mea-
surement method even as the same polymer name. Hence,
the system take in a lot of information about the monomer,
Simulation Database the polymer chain, the polymer aggregate and the polymer
materials to treat the characteristic of polymer sample (the
Supplement Polymer DB
parameters experiment sample) cyclopaedically.
•Molecular Polymer The monomer information includes a name, a molecular
Simulation Dictionary
Simulation •Properties formula, a structural drawing. The polymer information in-
Tools Estimation Monomer Polymerization Property
• Reactivity DB Information Data clude a structural unit (name, chemical structure, molecular
predication weight), the primary structure (stereoregularity, molecular
• Etc.

Calculated Structure NMR weight). And there is the polymerization information that
Results Valididation Information DB
is information to link the polymer chain to the monomer.
The polymer aggregate include information of higher-order
Figure 2. Structure of PoLyInfo. structure (information of polymer processing and crystal
architecture) and properties information. Moreover, the poly-
mer materials have materials name and commodity name.
Therefore, PoLyInfo is possible to analyze above informa-
In the simulation part, an application estimate properties tion for building data model.
of data not registered and virtual materials using real data
in database. In the properties estimation, there is a group B. Collecting of polymer information
contribution method in one of the techniques for properties The data source of PoLyInfo is the papers which registered
estimation from the chemical structure of constitutional re- Chemical Abstracts Service (CAS). We select papers to
peating unit (CRU) of polymer. We use properties estimation register with PoLyInfo among the following view points.
of Van Krevelen[9] in PoLyInfo and properties of prediction
4 International Union of Pure and Applied Chemistry(IUPAC)-compliant
object are shown as a relational expression of some factors.
name of structural foundation, a name of materials foundationCa popular
It calculates as sum of contribution from the atom group nameCabbreviated expressions and so on.
(atomic group parameter) included constitutional repeating 5 A polymer has the classifications such as a homopolymer, a copolymer,
unit (CRU). This method is possible to use database correlate a polymer blend. On account of the space, we omit it about the detailed
with chemical structure and properties value with effect contents.
6 thermal properties, electric properties, mechanical properties and so on
because the relational expression and the atomic group

23
 Table II Paper ID, Sheet composer, etc.
Paper Main Information
S CIENTIFIC JOURNALS WITH MANY PAPERS REGISTERED WITH K Sheet 1 ID
P O LY I NFO .
Polymer name, Composition unit,
L Sheet Chemical structure, Feature of Polymer Polymer
sample, etc. 1-1 1-2
No. Journal name Number of
paper Correspondence monomer,
Polymerization procedure, primary structure primary structure
1 Journal of Polymer Science 2,745 M Sheet
Stereo regularity, 1-1-1 1-1-2
Molar weight, etc.
2 Macromolecules 2,455
Conditions of molding, higher-order structure
3 Journal of Applied Polymer Science 2,230 N Sheet Sample configuration,
higher-order structure
Crystal architecture, etc. 1-1-1-1 1-1-1-2
4 Polymer 1,893
5 Makromolekulare Chemie 620
P Sheet Property Property Property Property Property Property
6 Polymer Preprints (American Chemical 609 ‛ᕈ
Property , Unit, 1-1-1-1- ‛ᕈ
1-1-1-1- ‛ᕈ
1-1-1-1- ‛ᕈ
1-1-1-2- ‛ᕈ
1-1-1-2- ‛ᕈ
1-1-1-2-
Society, Division of Polymer Chemistry) Measurement 1110-1 1-1-1- 4110-1
1-1-1- 9160-1
1-1-1- 1110-1
1-1-1- 4110-1
1-1-1- 9160-1
1-1-1-
7 Synthetic Metals 602 method, 1110-2 4110-2 9160-2 1110-2 4110-2 9160-2
Measurement
8 European Polymer Journal 541 condition, etc.
9 Polymer Engineering and Science 421
Annual Technical Conference Figure 3. Outline of Data Sheet.
10 418
- Society of Plastics Engineers
11 Journal of Macromolecular Science 370
Table IV
12 Polymer Journal 322 C ATEGORY OF PROPERTIES .
13 Journal of Chemical Physics 319
14 Journal of Physical Chemistry 309 Physical property Flexural property
15 Macromolecular Chemistry and Physics 290
Optical property Compression characteristic
16 Langmuir 281
17 Polymer International 275 Thermal property Creep characteristic
18 Journal of the American Chemical Society 269 Electrical property Heat characteristic
19 Journal of Applied Physics 251
20 Proceedings of SPIE 251 Physicochemical property Impact strength
Dilute solution property Hardness
Rheological property Heat resistance
Table III and Combustion
N UMBER OF OPEN DATA IN P O LY I NFO . Tensile property Other property
Shear property
Homopolymers 10,270
Copolymers 3,364
Polymer Blends 1,097
Composites 1,505 adopt hierarchical data structure because the scientific jour-
Monomers 16,390 nal for the collection is described the properties of one or
Property points 220,189 more polymer samples.
Literature data 12,843 K sheet in figure 3 are described basic information as the
sheet composer and paper ID that polymer samples are listed
in7 . We make a plural L, M and N sheet depending on the
number of the polymer and the structure if plural polymer
samples are listed in a paper8 . Moreover, the physicality
– The chemical structure is clear. value of each polymer sample is registered with the P
– The actual value includes various properties. sheet. And we make a plural P sheet like other seats if
– The method for measurement and measurement condition the measurement method and its condition of experiment
are listed. are plural in the same polymer sample. The P sheet is
We also register the calculation value using actual measure- divided into plural seats depending on properties because the
ment value (derived figure) and do not register citatory value. properties information that the sheet composer can register
The specialists in polymer do screening for 2,800 CAS for one polymer sample is more than 100 items as thermal
papers per year manually and select polymer sample from properties, electric properties and mechanical properties in
700 papers per year. They write properties information in Table IV.
data sheet and register with PoLyInfo (actually PostgreSQL).
The scientific journal with many papers registered with C. Polymer Dictionary
PoLyInfo is shown in table II. And brief overview of registry For method of registration polymer information, we man-
data is shown in table III. age polymer data using a compound registry system for
One polymer sample is comprised of data sheets to have polymer which we call it polymer dictionary and we made.
five hierarchies where a top is paper information in Figure 3. We assort and identify polymer using chemical structure of
As for the polymer data, there is difference in primary
7 We manage the bibliography information such as the author name of
structure, the highly advanced structure about the same
papers, the name of journal (magazine) at the other table.
polymer class. And the value of properties data change by a 8 The information about the polymer is registered with the L sheet. And
difference of method for measurement and the measurement the detail information of the name and the structure are registered with
condition in the same polymer sample. Hence, PoLyinfo polymer dictionary.

24
 D. Data structure for polymer search
A database to use in PoLyInfo consists of data sheet
information and polymer dictionary mentioned in previous
section, journal information and indexes for efficient search.
The data of the L, M and N sheet are comprised of
a table of one each, as shown in Figure 6. Each table
has the information about the characteristic, sheet ID and
Figure 4. Isomers of C4 H10 O.
polymer ID. And P sheet is divided into about 60 tables
every properties as P3110, P7110. P sheet also has properties
information, sheet ID and polymer ID like other tables.
The sheet ID succeeds to the sheet ID of the high rank
&GUETKRVKXG TWNG hierarchy. First, paper ID becomes the K sheet ID. Next, the
$CUKE 'NGOGPV 0WODGTEQWRNKPI FKTGEVKQP].GHV LQKPV KPHQTOCVKQP_ ]4KIJV LQKPV
KPHQTOCVKQP_ ]$TCPEJ LQKPV KPHQTOCVKQP_ $TCPEJ HQTOCV L sheet ID added serial numbers of L to the K sheet ID. In a
'ZCORNG  RQN[UV[TGPG similar way, M, N and P sheet ID added serial numbers of it
RNCKP P]# _]_]_ P]_
P]_]#_
to the sheet ID of the upper level. We call ID of N sheet in
Figure 6 sample ID (for example, 29429-16-1-1 and 00081-
'ZCORNG  RQN[ GVJ[NGPG QZKFG
1-1), we use this ID for the presentation such as lists of
RNCKP P]# _]_ P]_]_ search results in PoLyInfo.
P]_]#_
The polymer dictionary has polymer master table which
is comprised of polymer name, polymer group, structural
Figure 5. Structural Format of PoLyInfo.
data and compositional formula and monomer master table
which is described monomer information. Moreover, it has
polymerization path table to manage the linkage of the
polymer and the monomer and blend master table which
constitutional unit in polymer dictionary. The materials of is comprised of plural polymer. For example, ID 75227-3
constitutional formula are different even if a composition of L sheet in Figure 6 is comprised of plural polymer and
type is the same. In general, chemical physical properties blend ID is given. It is possible to know sample information
are different. We call such a thing a constitution isomer. For of this blend to search blend master table using this blend
example, Figure 4 shows a part of the structural isomer of ID. Polymer master table has polymer ID and there is
C4 H10 O9 . The structure is different by a place with −OH correspondence relation to polymer ID that is maintained
though all compositional formulas are same10 . The chemical on each seat.
structure of the constitutional unit express in combination of In addition, there is the table that paper information
atomic group which constitutes the chemical structure of the recording the polymer sample in PoLyInfo and this table
constitutional unit. The polymer name is based on IUPAC include the author information. PoLyInfo has index table
nomenclature system and sheet composer give a structural consisting of sample ID, polymer ID and the minimum and
foundation name and a raw materials foundation name[1], the maximums of each properties for the efficient of the
[2]. A structure format of PoLyInfo is shown in Figure 5. polymer sample search using the value of properties.
The format that is recorded in the polymer dictionary E. The advantage of using hierarchical data structure
becomes uniformization by the identification algorithm of
polymer dictionary. And the polymer dictionary has com- The search for the upper level sheet is possible to delete
positional formula with the structural unit base, chemical the most right value of current sheet ID and to do exact
formula weight and the polymer name like structural foun- match search because each sheet ID includes the ID of
dation name in others of the chemical structure information. the upper level sheet. The search for the lower level sheet
Therefore, it is possible to search based on various names is possible to do prefix search using current sheet ID. It
and chemical structures in polymer search feature to describe can anticipate a speedup of the search time because prefix
section IV. search processing is fast in generically. For example, if the
users want to know correspondence monomer and molecular
Early versions of PoLyInfo treat homopolymer only. Now, weight of sample ID 00081-1-1-1 (N sheet ID) in Figure 6,
PoLyInfo treat also co-polymer and polymer blend by the the users search ID 00081-1-1 in M sheet table. And the
expansion of the recent polymer dictionary. search about the properties of the polymer sample should
Moreover, PoLyInfo can manage blend dictionary, search ID to begin with 00081-1-1-1 for each P sheet.
monomer dictionary, polymerization dictionary and the data For example, the users understand that the polymer ID of
of journal information in others of the polymer dictionary. 00081-1-1 -1 is P01002 in N sheet table and get polymer
Herewith, it is possible to manage various information information of compositional formula and name to search
uniquely. polymer master tables because each sheet has polymer ID.
About the search of properties as rolling temperature of
the glass and electric permittivity, the users get sample ID
9 We show the example of the monomer daringly to facilitate understand- quickly using index tables.
ing. Polymer search uses polymer master table at first. Polymer
10 We call upper left on the Figure 4 n-butyl alcohol, left lower sec-butyl
alcohol, upper right tert-butyl alcohol and right lower isobutyl alcohol
master table has structural foundation name, raw materials

25
 WĂƉĞƌŝŶĨŽƌŵĂƚŝŽŶƚĂďůĞ;ƐƵďƐƚŝƚƵƚŝŽŶŽĨ<ƐŚĞĞƚ ƚĂďůĞͿ D ƐŚĞĞƚƚĂďůĞ
ID Polymer ID Authors’ name 䊶䊶䊶 Paper No. – L sheet No. 䊶䊶䊶 ID Polymer ID Feature 1 䊶䊶䊶
00081 P010001;P010002 S. Otsuka 00081-1;00081-2 28689-3-1 P373023 1
00109 P060014 I. Kuwajima 00109-1 28689-4-1 P373024 2
00078 P074244;P074254 M. Ymazaki 00078-1;00078-2 00081-1-1 P01002 1

> ƐŚĞĞƚƚĂďůĞ E ƐŚĞĞƚƚĂďůĞ

ID Polymer ID Blend ID Feature 1 䊶䊶䊶 ID Polymer ID Feature䋱 䊶䊶䊶
28689-3 P373023 1 29429-16-1-1 P903336 1
00081-1 P010002 2 75227-4-1-1 P900478 2
P020001//
75227-3 P044464 BD000189 1 00081-1-1-1 P01002 1

W ƐŚĞĞƚƚĂďůĞ
WϯϭϭϬ ƚĂďůĞ WϲϭϮϬ ƚĂďůĞ
ID Polymer ID Feature䋱 䊶䊶䊶 ID Polymer ID Feature 䋱 䊶䊶䊶
75227-1-1-1-3110-1 P044464 1 15108-2-1-1-6120-1 P0462313 1
14308-7-1-1-3110-1 P020001 2 00557-1-1-1-6120-1 P070317 3
00081-1-1-1-3110-1 P01002 1 00081-1-1-1-6120-1 P01002 2

Figure 6. Example of Data Sheet Tables.

foundation name, popular name and structural data. The A. Polymer search
users can get polymer ID corresponding to these names
in performing a text search for these columns. After, the The users can two type of search in PoLyInfo. One is basic
users can get the information about the polymer sample search and the other is advanced search. The screenshot
corresponding to these names. For monomer search, the of basic search is shown in Figure 7. Search items of
users get monomer ID using monomer master table and get basic search are polymer type, polymer name, polymer
monomer ID to search master table of polymerization path classification, materials type12 , polymer properties and paper
using the monomer ID. Polymer search, polymer structural information.
search and monomer search are realized by searching these For basic search of polymer name, the users can choice
tables. an element by select menu other than text search. For basic
The implementation of PoLyInfo avoid join operator that search of polymer properties, the users can decide the range
is high load processing in SQL because the data structure of the value of properties. For basic paper search, the users
of PoLyInfo is hierarchic. The search time is slow even if can specify journal name, author’s name and publication
making index for searches to each ID column because each year. It is possible to ‘ANDf search basically when the users
ID is resisted as character string. And it may take time by choice plural items.
the presentation of search results because a case becoming For advanced search, the users can specify various fields
the partial search occurs in polymer and monomer search. as average molecular weight, sample configuration, addition
Now, the enrollment of the polymer master table is around agent and crystallinity degree. And the number of properties
20,000 and we replaced a database system from ObjectStore items is three kinds.
to PostgreSQL. When users perform a complicated search,
results are displayed in one or two minutes. Search time B. Polymer structural search
may increase in future when amount of data growth. We
recognize that it is necessary to improve it in this connection. Polymer structural search is search feature to specify sub-
ject to the chemical structure of the structural. The method of
The sheet composers create the data sheet and the polymer
structural search are two way. One is easy structural search
dictionary using Microsoft Excel. The update work of the
and the other is advanced structural search. The users search
database and the rebuilding work of the index table are batch
can specify the kind and the number of the atomic group of
processing11 .
the structural using more than three hundred base element
IV. P OLYMER SEARCH FEATURE in the easy structural search. The advanced structural search
can specify bonding state of chemical structure using Java
The search feature of PoLyInfo includes three kinds applet. The screenshot of easy structural search is shown in
(polymer search, polymer structural search and monomer Figure 8 and advanced structural search is shown in Figure 9.
search).
12 In addition to this, there are polymer, composite and compound. On
11 These works are around several times in a year. account of the space, we omit it about the detailed contents.

26
 Figure 9. Polymer Structural Search using Original Tools.

Figure 7. Easy Plymer Search.

WŽůLJŵĞƌ
ƐĞĂƌĐŚ ^ĂŵƉůĞ
WŽůLJŵĞƌůŝƐƚ ^ĂŵƉůĞůŝƐƚ
ŝŶĨŽƌŵĂƚŝŽŶ
^ƚƌƵĐƚƵƌĞ
ƐĞĂƌĐŚ
ŽŶƐƚŝƚƵƚŝŽŶĂůhŶŝƚ
EDZĂƚĂďĂƐĞ
,ŝƐƚŽŐƌĂŵ ŝŶĨŽƌŵĂƚŝŽŶ

>ŝƐƚŽĨŵŽŶŽŵĞƌ >ŝƐƚŽĨƉŽůLJŵĞƌ

DŽŶŽŵĞƌ
DŽŶŽŵĞƌůŝƐƚ DŽŶŽŵĞƌŝŶĨŽƌŵĂƚŝŽŶ
ƐĞĂƌĐŚ

Figure 10. Transition of User Behavior.

formula and chemical structure and the users can move to

List of monomer to show the monomer of candidate raw ma-
terials and polymerization classification, co-polymer which
assumes the polymer concerned a structural unit, polymer
Figure 8. Polymer Structural Search using Atomic Selection.
blend and NMR(Nuclear Magnetic Resonance) information.
13
. In the monomer search, monomer information displaying
all information of the monomer is displayed via Monomer
list to show results list.
C. Monomer search
V. U SAGE OF OUR SYSTEM
The monomer search which is polymer raw materials
is possible to search to specify monomer name, registry A. Access trend of MatNavi
number of various chemical compound, molecular formula First, we describe access trend of MatNavi which is the
and molar weight. portal site about a materials database including PoLyInfo.
D. Search result and transition of usersf behavior User’s registration is necessary for using each database in
MatNavi. The users can use all databases including the basic
The users can next search and show details easily using crystal structures database, diffusion database and so on as
current search results in PoLyInfo. Transition of user be- well as PoLyInfo by performing user’s registration once.
havior after search is shown in Figure 10. First PoLyInfo There are 53,850 registered users, 141 countries and 16,730
displays the number of polymer, properties item and mea- organizations at March, 2011. So MatNavi is one of the most
surement value each structural unit which matched search famous materials database in the world. Figure 11 shows
condition (Polymer list of in Figure 10) if the users do an institute (organization) and area with much registration
polymer search. Next, it displays Sample list listing the users. The access trend of MatNavi is shown in Figure 12
polymer of the structural unit. Finally, it displays Sample and there are around one million accesses every month. And
information displaying all information including the proper- the access trend of a main database is shown in Figure 13.
ties of each polymers. The users can check crystal condition The Polymer database is the most access in MatNavi14 .
in Histogram which can move from polymer list.
And the users can show Constitutional Unit information 13 We get bonding state of atomic in detail using NMR.
(CU Information) to move from polymer list. The CU Infor- 14 InFigure 12 and 13, the access means the number of the hits (the
mation displays names of the structural unit, compositional simplest count method including the access of images files).

27
 500,000

450,000

400,000

ss
ec 350,000

ca
fo
300,000

erb
250,000

m
u
200,000

N 150,000

100,000

50,000

60
r-p 06- 60
-g
60 60 70
t-c c-e -be
70
r- 07- 70
-
70 70 80
t-c c-e -be
80
r-p 08- 80
-g 08t- 08- 90
-b
unJ unJ uAg unJ cO ce
A uA O D F Ap O D F A uA D eF

Figure 14. Transition of a Number of Accesses for MatNavi.

Figure 11. Users Locations.

ϭϲϬ͕ϬϬϬ

ϭϰϬ͕ϬϬϬ

ϭϮϬ͕ϬϬϬ
ƐƐĞ
ĐĐ ϭϬϬ͕ϬϬϬ
ĂĨ
sit Žƌ ϴϬ͕ϬϬϬ
hf Ğď
or Ƶŵ ϲϬ͕ϬϬϬ
eb E
ϰϬ͕ϬϬϬ
m
u
N ϮϬ͕ϬϬϬ

Ϭ
ϲϬ ϲϬ ϲϬ ϲϬ ϲϬ ϳϬ ϳϬ ϳϬ ϳϬ ϳϬ ϳϬ ϴϬ ϴϬ ϴϬ ϴϬ ϴϬ ϴϬ ϵϬ
Ͳƌ ͲŶ ͲŐ ƚͲĐ ͲĐ Ͳď Ͳƌ ͲŶ ͲŐ ƚͲĐ ͲĐ Ͳď Ͳƌ ͲŶ ͲŐ Ͳ ͲĐ Ͳď
Ɖ Ƶ: Ƶ Ğ Ğ& Ɖ Ƶ: Ƶ Ğ Ğ& Ɖ Ƶ: Ƶ Đƚ Ğ Ğ&
  K    K    K 

Figure 12. Transition of a Number of Hits for MatNavi. Figure 15. Transition of a Number of Accesses for PoLyInfo.

㪋㪌㪇㪃㪇㪇㪇 㜞ಽሶ
Polymer
㪋㪇㪇㪃㪇㪇㪇 ⚿᥏ၮ␆
Crystal increases year by year although the number of the access
sit 㪊㪌㪇㪃㪇㪇㪇 ᭴ㅧ᧚ᢱ
Structural spikes by each month. The number of the access in August
hf ᜛ᢔ
Diffusion is low because many users have a vacation.
or 㪊㪇㪇㪃㪇㪇㪇
eb 㪉㪌㪇㪃㪇㪇㪇 Transition of a number of accesses for PoLyInfo is shown
m
u 㪉㪇㪇㪃㪇㪇㪇 in Figure 15. The average of number of the access is 40-
N 50 thousands per month although the number of the access
㪈㪌㪇㪃㪇㪇㪇
㪈㪇㪇㪃㪇㪇㪇
spikes by each month. The number of the access at October,
㪌㪇㪃㪇㪇㪇
2006 is high in order to attack by the automatic collection
㪇
program.
C. A difference of the database use situation by the user
Figure 13. Transition of a Number of Hits for Main Databases.
We analyze the use situation of database in MatNavi by
a Japanese university and the Japanese company because
web access logs include access IP information. The term
B. Study of access trend of analysis is from April, 2008 to march, 2009 and we
We use Apache for a Web server of MatNavi15 . Web added up the number of the access each day. The results
access logs are divided into every database because we of PoLyInfo access is shown in Figure 16 and the results of
install the Web server in every each database16 . We analyze basic crystal structures database (Pauling file) is shown in
web access logs from April, 2006 to March, 2009. Figure 17. As results, the number of access from Japanese
At first, we unified all web access logs as a preliminary university in basic crystal structures database is high. And
arrangement to analyze the logs. And we cleaned it using the number of access from Japanese company in PoLyInfo
following steps. is high. We know that the Japanese company performs a
polymer research lively.
– Delete image files and css files.
– Delete web crawler access. VI. C ONCLUSION AND FUTURE WORK
– Delete access to error pages.
In this paper, we describe system structure, search method
Transition of a number of accesses for MatNavi included and usage of PoLyInfo (and MatNavi).
all database is shown in Figure 14. The number of the access Now, it is possible to treat homo-polymer, co-polymer and
15 We remove local access from access logs.
polymer blend in PoLyInfo. However, it is not possible to
16 Afterward, we assume that the databases have web server function treat branched polymer like dendrimer and hyperbranched

28
 Figure 16. Difference of Access Trend between Universities and Companies (PoLyInfo).

Figure 17. Difference of Access Trend between Universities and Companies (Basic crystal Structures Database).

polymer. We level up the polymer dictionary for collect and [4] T. Maeda. Construction and use of polymer database.
express all polymers that structure is clear. POLYMERS, 48(5):320–323, 1999.
And the work to collect data from papers and to make [5] T. Maeda. Role of database to polymer research. POLY-
each dictionary is carried out with human power. From the MERS, 51(6):442–453, 2002.
[6] K. Nagasaka. Construction of polymer physical properties
viewpoint of budget, simplification of this work is one of the database and physical properties prediction approaches. CI-
important problems. In the future, we modify our system CSJ Bulletin, 22(3):38–43, 2004.
about growth of properties, speedup of search and user- [7] J. Rumbaugh, M. Blaha, W. Premerlani, F. Eddy, and
friendly interface. W. Lorensen. Object-Oriented Modeling and Design. Pren-
tice Hall, 1991.
ACKNOWLEDGMENTS [8] K. Shiono, T. Maeda, J. Kawashima, S. Iimuro, K. Mashita,
This work was supported by a Grant-in-Aid for Young K. Nagasaka, Y. Takaeda, Y. Yoshinari, and K. Matsuda.
Polymer database ”polyinfo”. JCPE Journal, 13(4):225–234,
Scientists (B) from Japan Society for the Promotion of 2001.
Science (JSPS) under Grant No. 21700124. [9] D. van Krevelen. Properties of Polymers, 3rd edition.
Elsevier, 1990.
R EFERENCES [10] M. Yamazaki. Current status of polymer database (polyinfo).
[1] P. A. Chem. Nomenclature of regular single-strand organic CICSJ Bulletin, 25(4):103–107, 2007.
polymers. Rules Approved 1975, 48:373–385, 1976.
[2] J. Committee of Polymer Nomenclature, The Society of
Polymer Science. Terminology and Nomenclature of Polymer.
Kodansha, 2007.
[3] I. Kuwajima, M. Kumazaki, K. Mashita, J. Hosoya, and
M. Yamazaki. Current status and future development of
”polymer database (polyinfo)”. In The 2nd Asian Materials
Database Symposium (AMDS 2010), pages 3E–01, March
2010.

29