The ADS services can be accessed through various interfaces. Some of these interfaces use WWW based forms, others allow direct access to the database and search system through Application Program Interfaces (APIs). This section describes the various interfaces and their use, as well as the returned results.
a. User interface
The main query forms (Figs. 1, 2, 3) provide access to the different abstract databases. These forms are generated on demand by the ADS software. This allows the software to check the user identification through the HTTP (HyperText Transfer Protocol) cookie mechanism (see Sect. 3), so that the software can return a customized query form if one has been defined by the user. It also adapts parts of the form according to the capabilities of the user's web browser.
![\begin{figure}\includegraphics[width=10cm]{DS1781F1.eps}\end{figure}](img1.gif) |
Figure 1: The ADS Abstract Service query form provides the capability to query the database by authors, object names, title and text words |
![\begin{figure}\includegraphics[width=10cm]{DS1781F2.eps}\end{figure}](img2.gif) |
Figure 2: The Filter section of the query form allows selection of references that have specific properties |
The query form allows the user to specify search terms in different fields. The input parameters in each query field can be combined in different ways, as can the results obtained from the different fields (Fig. 1). The user can specify how the results are combined through settings on the query form (Fig. 3). The combined results can then be filtered according to various criteria (Fig. 2).
The database can be queried for author names, astronomical objects names, title words, and words in the abstract text. References can be selected according to the publication date. The author name, title, and text fields are case insensitive. The object field is case sensitive when the IAU (International Astronomical Union) Circulars (IAUC) object name database is searched, since the IAU object names are case sensitive. In the author and object name fields, the form expects one search term per line since the terms can contain blanks. In the title and text fields line breaks are not significant.
The author names are indexed by last name and by a combination of last name and first initial, separated by a comma. To account for differences in the spelling of the same author name, the search system contains a list of author names that are spelled differently but are in fact names of the same author. This allows for automatically retrieving all versions of common spelling differences. This is useful for instance for German umlaut spelled as Muller and Mueller, or variations in the transliteration of names from non-English alphabets like Cyrillic. An example of such an entry in the author synonym list is:
AFANASJEV, V AFANAS'EV, V AFANAS'IEV, V AFANASEV, V AFANASYEV, V AFANS'IEV, V AFANSEV, VWithout this synonym replacement capability, author searches would obviously be much less effective. On user request we also include name changes (e.g. due to marriage) in the author synonym list. Combinations of search results within the author field use "OR'', "AND'', or simple logic (see below), depending on user selection.
Author names are quite often spelled differently in different publications. First names are sometimes spelled out, sometimes only first initials are given, and sometimes middle initials are left out. This makes it impossible to index all different spellings of a name together automatically.
To handle these different requirements, author names are indexed three times, once with the last name only, once with the last name and first initial, and once with the complete name as it is specified in the article.
To access these different indexes, we provide two user interfaces for author queries. The regular user interface allows the user to search for either a last name or a last name combined with the first initial. This allows for fairly discriminating author searches. It is a compromise between the need to discriminate between different authors, and the need to find all instances of a given author. It identifies all different versions of a given author quite reliably, but it indexes together different authors with the same first initial. For cases where this search method is not discriminating enough, we provide a second user interface to the index of the full names, which does not attempt to index different spellings of the same author together. When the user selects "Exact Author Search'' and specifies an author's last name or last name and first initial, a form is returned with all distinct full author names that match the specified name. The user then selects all the different spellings of the desired name and queries the database for articles that contain any one of these different versions of an author's name. For instance specifying:
Eichhorn, G
in the exact author name form returns the list:
EICHHORN, G. EICHHORN, GERHARD EICHHORN, GUENTHER EICHHORN, GUNTHER
Selecting the first, third, and fourth author name from that list will return all articles by the first author of this article. Any articles by the second author containing only the first initial will also be returned, but this is unavoidable.
This field allows the user to query different databases for references with different astronomical objects. The databases that provide object information are: SIMBAD (Set of Identifications, Measurements and Bibliographies for Astronomical Data) at the Centre des Données Astronomique de Strasbourg (CDS), France ([Egret et al. 1991]); the NASA Extragalactic Database (NED) at the Infrared Processing and Analysis Center (IPAC), Jet Propulsion Laboratory (JPL), Pasadena, CA ([Madore et al. 1992]); the IAU Circulars (IAUC) and the Minor Planet Electronic Circulars (MPEC), both provided by the Central Bureau for Astronomical Telegrams (CBAT) at the Harvard-Smithsonian Center for Astrophysics in Cambridge, MA ([Marsden 1980]); and a database with objects from publications from the Lunar and Planetary Institute (LPI) in Houston (mainly Lunar sample numbers and meteorite names). The user can select which of these databases should be queried. If more than one database is searched, the results of these queries are merged. The LPI database does not have any entries in common with the other databases. The SIMBAD, NED, and IAUC databases sometimes have information about the same objects.
These fields query for words in the titles of articles or books, and in the abstracts of articles or descriptions of books respectively. The words from the title of each reference are also indexed in the text field so they will be found through either a title or a text search. Before querying the database the input in these fields is processed as follows:
1. Apply translation rules. This step merges common expressions into a single word so that they are searched as one expression. Regular expression matching is used to convert the input into a standard format that is used to search the database. For instance M 31 (with a space) is translated to M31 (without a space) for searching as one search term. In order to make this general translation, a regular expression matching and substitution is performed that translates all instances of an "M'' followed by one or more spaces or a hyphen followed by a number into "M'' directly followed by the number. Other translation rules include the conversion of NGC 1234 to NGC1234, contractions of T Tauri, Be Star, Shoemaker Levy, and several others (see ARCHITECTURE).
2. Remove punctuation. In this step all non-alphanumeric characters are removed, unless they are significant (for instance symbols used in the simple logic (see below), "+'' and "-'' before numbers, or ".'' within numbers).
3. Translate to uppercase. All information in the index files is in uppercase, except for object names from the IAU Circulars.
4. Remove kill words. This step removes all non-significant words. This includes words like "and'', "although'', "available'', etc. (for more details see ARCHITECTURE).
In the title and text fields, searching for phrases can be specified by enclosing several words in either single or double quotes, or concatenating them with periods (".'') or hyphens ("-''). All these accomplish the same goal of searching the database for references that contain specified sequences of words. The database is indexed for two-word phrases in addition to single words. Phrases with more than two words are treated as a search for sets of two-word phrases containing the first and second word in the first phrase, the second and third word in the second phrase, etc.
b. Searching
After the search terms are pre-processed, the databases of the different fields are searched for the resulting list of words, the results are combined according to the selected combination rules, and the resulting score is calculated according to the selected scoring criteria. These combination rules provide the means for improving the selectivity of a query.
The database is searched for the specified words as well as for words that are synonymous with the specified term. One crucial part to successful searches in a free text search system is the ability to not only find words exactly as specified, but also similar words. This starts with simply finding singular and plural forms of a word, but then needs to be extended to different words with the same meaning in the normal usage of words in a particular field of science. In Astronomy for instance "spectrograph'' and "spectroscope'' have basically the same meaning and both need to be found when one of these words is specified in the query. Even further reaching, more discipline-specific synonyms are necessary for efficient searches such as "metallicity'' and "abundance'' which have the same meaning in astronomical word usage. In order to exhaustively search the database for a given term, it is important to search for all synonyms of a given word. The list of synonyms was developed manually by going through the list of words in the database and grouping them according to similar meanings. This synonym list is a very important part of the ADS search system and is constantly being improved (see ARCHITECTURE).
The list of synonyms also contains non-English words associated with their English translations. These words came from non-English reference titles that we included in the database. This allows searches with either the English or non-English words to find references with either the English word or the non-English translation. We are in the process of extending this capability by including translations of most of the words in our database into several languages (German, French, Italian, Spanish). This will allow our users to phrase queries in any of these languages. We expect to complete this project sometime in 2000.
By default a search will return references that contain the search word or any of its synonyms. The user can choose to disable this feature if for some reason a specific word needs to be found. The synonym replacement can be turned off completely for a field in the "Settings'' section of the query form. This can be used to find a rare word that is a synonym of a much more frequent word, for instance if you want to look for references to "dateline'', which is a synonym to "date''. Synonym replacement can also be enabled or disabled for individual words by prefixing a word with "='' to force an exact match without synonym replacement. When synonym replacement is disabled for a field, it can be turned on for a particular word by prefixing it with "#''.
There are four different types of combinations of results for searches within a field possible.
1. OR 2. AND 3. Simple logic 4. Full boolean logic
1. Combination by "OR'':
The resulting list contains all references that contain at least one
of the search terms.
2. Combination by "AND'': The resulting list has only references that contain every one of the search terms.
3. Combination by simple logic: The default combination in this logic is by "OR''. Individual terms can be either required for selection by prefixing them with a "+'', or can be selected against by prefixing them with a "-''. In the latter case only references that do not contain the search term are returned. If any of the terms in the search is prefixed by a "+'', any other word without a prefix does not influence the resulting list of references. However, the final score (see below) for each reference will depend on whether the other search terms are present.
4. Combination by full boolean logic: In this setting, the user specifies a boolean expression containing the search terms and the boolean operators "and'', "or'', and "not'', as well as parentheses for grouping. A boolean expression could for instance look like:
(pulsar or "neutron star'') and ("red shift'' distance) and not 1987A.
This expression searches for references that contain either the word pulsar or the phrase "neutron star'' and either the phrase "red shift'' or the word distance ("or'' being the default), but not the word 1987A.
In the settings part of the query form, the user can specify fields that will be required for selection. If a field is selected as "Required for Selection'' only references that were selected in the search specified in that field will be returned. If one field is selected as "Required for Selection'', the searches in fields that are not set as "Required for Selection'' do not influence the resulting list, but they influence the final score.
c. Scoring
The list of references resulting from a query is sorted according to a "score'' for each reference. This score is calculated according to how many of the search items were matched. The user has the choice between two scoring algorithms:
1. proportional scoring 2. weighted scoringThese scoring algorithms have been analyzed by [Salton & McGill (1983)].
In proportional scoring, the score is directly proportional to the number of terms found in the reference. In weighted scoring, the score is proportional to the inverse logarithm of the frequency of the matched word. This weighting gives higher scores for words that are less frequent in the database and therefore presumably more important indicators of the relevance of a match. In the settings section of the query form the user can select which type of scoring should be used for each query field separately. The default setting for title and text searches is the weighted scoring. For author searches proportional scoring is the default. Once the score for each query field is calculated, the scores are normalized so that a reference that matches all words in a field receives a score of 1.
The normalized scores from the different fields are then combined to calculate a total score. Again the result is normalized so that a reference that matches all words in each query field has a score of 1. The user can influence this combining of scores from the different search fields by assigning weights to the different fields. This allows the user to put more emphasis in the selection process on, for instance, the object field by assigning a higher weight to that field. Another use of the weight field is to select against a field. For instance specifying an object name and an author name and selecting a negative weight to the author field will select articles about that object that were not written by the specified author.
The relative weights for the different search fields can be set by the user. The ADS provides default weights as follows:
Authors: 1.0 Objects: 1.0 Title: 0.3 Text: 3.0
These default weights were determined on theoretical grounds, combined with trial and error experimentation. We used different search inputs from known research fields and different weights and ranked the resulting lists according to how well they represented articles from these research fields. The weights listed above gave the best results.
d. Filtering of selected references
The selected references can be filtered according to different criteria (see Sect. 4.5) in order to reduce the number of returned references. The user can select references according to their entry date in the database, a minimum score (see above), the journal they are published in, whether they have pointers to selected external data sources, or whether they belong to one or more of several groups of references. This allows a user for instance to select only references from refereed journals or from one particular journal by specifying its abbreviation. It also allows a user to select only references that have links to external data sets, on-line articles, or that have been scanned and are available through the ADS Article Service.
e. Display of search results
The ADS system returns different amounts of information about a reference, depending on what the user request was. This section describes the different reference formats.
The list of references returned from a query is displayed in a tabular format. The returned references are sorted by score first. For equal scores, the references are sorted by publication date with the latest publications displayed first.
A typical reference display is shown in Fig. 4. The fields in such a reference are shown in Fig. 5. They are as follows:
1. Bibliographic Code: This code identifies the reference uniquely (see DATA and [Schmitz et al. 1995]). Two important properties of these codes are that they can be generated from a regular journal reference, and that they are human readable and can be understood and interpreted.
2. Score: The score is determined during the search according to how well each reference fits the query.
3. Date: The publication date of the reference is displayed as mm/yyyy.
4. Links: The links are an extremely important aspect of the ADS. They provide access to information correlated with the article. Table 1 shows the links that we currently provide when available.
| Link | Resource | Description |
| A | Abstract | Full abstract of the article. These abstracts come from different sources. |
| C | Citations | A list of articles that cite the current article. This list is not necessarily complete (see |
| "R'' References). | ||
| D | On-line Data | Links to on-line data at other data centers. |
| E | Electronic Article | Links to the on-line version of the article. These on-line versions are in HTML format for |
| viewing on-screen, not for printing. a | ||
| F | Printable Article | Links to on-line articles in PDF or Postscript format for printing. a |
| G | Gif Images | Links to the images of scanned articles in the ADS Article Service. |
| I | Author Comments | Links to author supplied additional information (e.g. corrections, additional references, |
| links to data). | ||
| L | Library Entries | Links to entries in the Library of Congress on-line system. |
| M | Mail Order | Links to on-line document delivery systems at the publisher/owner of the article. |
| N | NED Objects | Access to lists of objects for the current article in the NED database. |
| O | Associated Articles | A list of articles that are associated with the current article. These can be errata or other |
| articles in a series. | ||
| P | Planetary Data System | Links to datasets at the Planetary Data System. |
| R | References | A list of articles referred to in the current article. For older articles these lists are not nec- |
| essarily complete, they contain only references to articles that are in the ADS database. | ||
| For articles that are on-line in electronic form, the "R" link points to the on-line reference | ||
| list, and therefore the complete list of references in that article. a | ||
| S | SIMBAD Objects | Access to lists of objects for the current article in the SIMBAD database. |
| T | Table of Contents | Links to the list of articles in a books or proceedings volume. |
| X | Planetary Nebulae | Links to datasets in the Galactic Planetary Nebulae Database. |
A more detailed description of resources in the ADS that these links point to is provided in DATA.
Some of these links (for instance the "D'' links) can point to more than one external information provider. In such cases the link points to a page that lists the available choices of data sources. The user can then select the more convenient site for that resource, depending on the connectivity between the user site and the data site.
5. Authors: This is the list of authors for the reference. Generally these lists are complete. For some of the older abstracts that we received from NASA/STI, the author lists were truncated at 5 or 10 authors, but every effort has been made to correct these abbreviated author lists (see DATA).
6. Title: The complete title of the reference.
The reference lists are returned as forms if table display is selected (see Sect. 3). The user can select some or all of the references from that list to be returned in any one of several formats:
i. HTML format: The HTML (HyperText Markup Language) format is for screen viewing of the formatted record.
ii. Portable Format: This is the format that the ADS uses internally and for exchanging references with other data centers. A description of this format is available on-line at:
![\begin{figure}\includegraphics[width=10cm]{DS1781F3.eps}\end{figure}](img3.gif)
![\begin{figure}\includegraphics[width=10cm]{DS1781F4.eps}\end{figure}](img4.gif)
![\begin{figure}
\includegraphics[width=7.5cm,clip]{DS1781F5.eps}\end{figure}](img5.gif){kind=small}
http://adsabs.harvard.edu/abs_doc/
abstract_format.html
iii. BibTeX format: This is a standard format that is used to build reference lists for TeX (a typesetting language especially suited for mathematical formulas) formatted articles.
iv. ASCII format: This is a straight ASCII text version of the abstract. All formatting is done with spaces, not with tabs.
v. User Specified Format: This allows the user to specify in which format to return the reference. The default format for this selection is the bibitem format from the AASTeX macro package. The user can specify an often used format string in the user preferences (see Sect. 3). This format string will then be used as the default in future queries.
The user can select whether to return the selected abstracts to the browser, a printer, a local file for storage, or email it to a specified address.
In addition to the information in the short reference list, the full abstract display (see Fig. 6) includes, where available, the journal information, author affiliations, language, objects, keywords, abstract category, comments, origin of the reference, a copyright notice, and the full abstract. It also includes all the links described above.
For abstracts that are displayed as a result of a search, the system will highlight all search terms that are present in the returned abstract. This makes it easy to locate the relevant parts in an abstract. Since the highlighting is somewhat resource intensive, it can be turned off in the user preference settings (see Sect. 3).
For convenience, the returned abstract contains links that allow the user to directly retrieve the BibTex or the custom formatted version of the abstract.
The full abstract display also includes a form that provides the capability to use selected information from the reference to build a new query to find similar abstracts. The query feedback mechanism makes in-depth literature searches quick and easy. The user can select which parts of the reference to use for the feedback query (e.g. authors, title, or abstract). The feedback query can either be executed directly, or be returned as a query form for further modification before executing it, for instance to change the publication date range or limit the search to specific journals. This query feedback mechanism is a very powerful means to do exhaustive literature searches and distinguishes the ADS system from most other search systems. A query feedback ranks the database against the record used for the feedback and sorts it according to how relevant each reference is to the search record. The query feedback can be done across databases. For instance a reference from the Astronomy database can be used as query feedback in the preprint database to see the latest work in the field of this article.
If the article for the current reference has been scanned and is available through the ADS Article Service (see below), printing options are available in the abstract display as well. These printing options allow the printing of the article without having to retrieve the article in the viewer first.
This part of the ADS provides access to the scanned images of articles. We have received permission from most astronomy journals to scan their volumes and make them available on-line free of charge. A more detailed description of these data is in DATA.
The most common access to the scanned articles is through the ADS Abstract Service via the "G''-links (see above). However they can also be accessed directly through the article query page by publication year and month or by volume and page at:
http://adsabs.harvard.edu/article_service.html
This form returns the specified article in the user specified format (see Sect. 3). If a page within an article is specified, and the single page display is selected, the specified page within the article is returned with the links to the other available pages as usual.
The article display normally shows the first page (Fig. 7) of an article at the selected resolution and quality (see Sect. 3). The user can select resolutions of 75, 100, or 150 dots per inch (dpi) and image qualities of 1, 2, 3, or 4 bits of greyscale per pixel. These gif images are produced on demand from the stored tiff images (see DATA). The default version of the gif images (100 dpi, 3 bit greyscale) is cached on disk. The cache of these gif images is managed to stay below a maximum size. Any time the size of the cached gif images exceeds the preset cache size, the gif images of pages that have not been accessed recently are deleted.
![\begin{figure}\includegraphics[width=10cm]{DS1781F7.eps}\end{figure}](img8.gif) |
Figure 7: The article display shows a gif image of the selected journal page with the resolution selected in the user preferences |
Below the page image on the returned page are links to every page of the article individually. This allows the user to directly access any page in the article. Wherever possible, plates that have been printed separately in the back of the journal volume have been bundled with the articles to which they belong for ease of access. The next part of the displayed document provides access to plates in that volume if the plates for this journal are separate from the articles. Another link retrieves the abstract for this article.
The next part of the page allows the printing of the article. If the browser works with HTTP persistent cookies (see Sect. 3), there is just one print button in that section with a selection to print either the whole paper or individual pages. This print button will print the article in the format that the user has specified in the user preferences. If the browser does not handle cookies, several of the more commonly used print options are made available here.
All possible printing options can be accessed through the next link called "More Article Retrieval Options''. This page allows the user to select all possible retrieval options. These include:
i. Postscript: Access to two resolutions is provided (200 dpi and 600 dpi). For compatibility with older printers, there is also an option to retrieve Postscript Level 1 files.
ii. PCL (Printer Control Language): This language is for printing on PCL printers such as the HP desk jets and compatibles.
iii. PDF (Portable Document Format): PDF can be viewed with the Adobe Acrobat reader ([Adobe, Inc]). From the Acrobat reader the article can be printed.
iv. TIFF (Tagged Image File Format): The original images can be retrieved for local storage. This would allow further processing like extraction of figures, or Optical Character Recognition (OCR) in order to translate the article into ASCII text.
v. Fax retrieval: Within the U.S.A., articles can be retrieved via fax at no cost. Again, the retrieval is greatly facilitated through the preferences setting capability. The preferences allow the user to store a fax number that will be used for the fax service.
vi. Email retrieval: Articles can be retrieved through email instead of through a WWW browser. MIME (Multipurpose Internet Mail Extension, [Vaudreuil 1992]) capable email systems should be able to send the retrieved images directly to the printer, to a file, or to a viewer, depending on what retrieval option was selected by the user.
For most of the retrieval options, the data can optionally be compressed before they are sent to the user. Unix compress and GNU gzip are supported compression algorithms.
Instead of displaying the first page of an article together with the other retrieval links, the user has the option (selected through the preferences system, see Sect. 3) to display thumbnails of all article pages simultaneously. This allows an overview of the whole article at once. One can easily find specific figures or sections within an article without having to download every page. This should be especially useful for users with slow connections to the Internet. Each thumbnail image ranges in size from only 700 bytes to 3000 bytes, depending on the user selected thumbnail image quality. The rest of this type of article page is the same as for the page-by-page display option.
There are several forms available to directly access references or articles and other relevant information. All abstract query forms return the short reference format as described above. One form allows access to references through bibliographic codes:
http://adsabs.harvard.edu/bib_abs.html
This form allows the user to retrieve abstracts by specifying directly a bibliographic code or the individual parts of a bibliographic (year, journal, volume, page). This can be very useful in retrieving references from article reference lists, since such reference lists generally contain enough information to build the bibliographic codes for the references. This form also accepts partial codes and returns all references that match the partial code. It accepts the wildcard character "?''. The "?'' wildcard stands for one character in the code. For partial codes that are shorter than 19 characters, matching is done on the first part of the bibliographic codes. For instance:
1989ApJ...341?...1 will retrieve the articles on page 1 of the ApJ (Astrophysical Journal) and ApJ Letters volume 341, regardless of the author name.
Another form allows access to the Tables of Contents (ToCs) of selected journals by month/year or volume:
http://adsabs.harvard.edu/toc_service.html
One option on that form is to retrieve the latest published issue of a particular journal. Access to the last volumes of a set of journals is also available through a page with a graphical display of selected journals' cover pages:
http://adsabs.harvard.edu/tocs.html
By clicking on a journal cover page either the last published volume of that journal or the last volume that the user has not yet read is returned, depending on the user preference settings (see Sect. 3). The information necessary for that service is stored with the user preferences in our internal user preferences database.
A customized ToC query page is available at:
http://adsabs.harvard.edu/custom_toc.html
It will display only icons for journals that have issues available that have not been read by the user. This allows a user to see at a glance which new issues for this set of journals have been published. The set of journals that is included in the customized ToC query page can be specified in the user preferences (see Sect. 3).
As mentioned in Sect. 2.1.1 and in ARCHITECTURE, one important aspect of the ADS search system is the list of synonyms. Sometimes it is important for our users to be able to see what words are in a particular synonym group to properly interpret the search results. Another question that is asked is what words are in the database and how often. The list query page (linked to the words "Authors'', "Title Words'', and "Text Words'' above the corresponding entry fields on the main query form) allows the user to find synonym groups and words in the database. The user can specify either a complete word in order to find its synonyms, or a partial word with wildcard characters to find all matching words in the database. When a word without wildcard characters is specified, the list query form returns all of its synonyms (if any).
To find words matching a given pattern, the users can specify a partial word with either or both of two wildcard characters. The question mark "?'' stands for any single character, the asterisk "*'' stands for zero or more characters (see Sect. 4.2.3). For a wildcard search, the list query form returns all words in the database that match the specified pattern, together with the frequencies of these words in the database.
The regular query forms search the complete database. The user can select the return of only specific journals in the "Filter'' section of the query form. In order to allow different journals to use the ADS system for searching their references, journal specific pages are available. The URL (Uniform Resource Locator) for an abstract search page for a specific journal is:
![\begin{figure}\includegraphics[width=10cm]{DS1781F6.eps}\end{figure}](img7.gif)
http://adsabs.harvard.edu/Journals/
search/bibstem
The page for retrieving scanned articles of a specific journal is:
http://adsabs.harvard.edu/Journals/
articles/bibstem
and the page for retrieving the tables of contents by volume or publication date is:
http://adsabs.harvard.edu/Journals/toc/bibstem
In each case, bibstem is the abbreviation for the selected journal. For instance:
http://adsabs.harvard.edu/Journals/search/ApJ
returns a query form for searching only references of the Astrophysical Journal. These forms are available for linking by anybody.
Both abstracts and articles can be accessed directly though HTML hyperlinks. The references are identified through the bibliographic codes (or bibcodes for short) mentioned above and described in detail in DATA. The syntax for such links to access abstracts is:
http://adsabs.harvard.edu/cgi-bin/
bib_query?bibcode=1989ApJ...342L..71R
Scanned articles can be accessed directly through links of the form:
http://adsabs.harvard.edu/cgi-bin/
article_query?bibcode=1989ApJ...342L..71R
These links will return the abstract or scanned article respectively for the specified bibliographic code. They are guaranteed to work in this form. You may see other URLs while you use the ADS. These are internal addresses that are not guaranteed to work in the future. They may change names or parameters. Please use only links of the form described above to directly access abstracts and articles.
Embedded queries can be used to build hyperlinks that return the results of a pre-formulated query. One frequently used example is a link that returns all articles written by a specific user. The syntax for such a link is:
<a href="http://adsabs.harvard.edu/cgi-bin/
abs_connect?return_req=no_params&
param1=val1¶m2=val2&...">...</a>
There are no spaces allowed in a URL. All blanks need to be encoded as "+''. The parameter return_req=no_params sets all the default settings. Individual settings can be changed by including the name of the specific setting and its value after the return_req=no_params parameter in the URL. A list of available parameters can be accessed at:
http://adsdoc.harvard.edu/cgi-bin/
get_field_names.pl
We try to make changes to parameters backward compatible, but that is not always possible. We encourage you to use this capability, but it is advisable to use only the more basic parameters.
This type of interface allows users to link to the ADS for a comprehensive list of references on a specific topic. Many users use this to provide an up-to-date publication list for themselves by encoding an author query into an embedded query.
The ADS database can be used by other systems to include ADS data in documents returned from that site. This allows programmers at other sites to dynamically include the latest available information in their pages. An example is the interface that SPIE (the International Society for Optical Engineering) provides to our database. It is available at:
http://www.spie.org/incite/
This site uses Perl scripts to query our database and format the results according to their conventions. These Perl scripts are available at:
http://adsabs.harvard.edu/www/adswww-lib/
The Perl scripts allow the programmer to specify all the regular parameters. The results are returned in Perl arrays. If you use these scripts, we would appreciate it if you would credit the ADS somewhere on your pages.
The ADS Abstract Service can be accessed through an email interface. This service is somewhat difficult to use since it involves an interface between two relatively incompatible interface paradigms. This makes describing it quite difficult as well. It is intended for users who do not have access to web browsers. If you have questions about how to use this access, please contact the ADS at ads@cfa.harvard.edu.
To get information about this capability, send email to:
adsquery@cfa.harvard.edu
with the word "help'' in the message body.
This interface accepts email messages with commands in the message body. The subject line is ignored. The commands that are available are:
1) help (see above) 2) action=URL
The second command allows a user to retrieve a document at the specified URL. Three qualifiers allow the user to specify what retrieval method to use, what format to return, and to which address to return the results:
a) method="method'' b) return="return-type'' c) address="e-mail-address''
a. "method'' is either "get'' or "post'' (without the quotes). This determines what kind of query will be executed. To retrieve a form for further queries, use the `get' method. To execute a forms query you need to know what type of query the server can handle. If you execute a forms query after retrieving the form through this service, the correct method line will already be in place. Default method is "get''.
b. "return-type'' is either "text'', "form'', or "raw'' (without the quotes). If text return is requested, only the text of the query result is returned, formatted as if viewed by a WWW browser. If form return is requested, the text of the result is returned as well as a template of the form that can again be executed with this service. If raw return is requested, the original document is returned without any processing. Default return is "form''. The capability to return MIME encoded files is in preparation.
c. "e-mail-address'' specifies to which e-mail address the result should be sent. This line is optional. If no address is specified, the result is sent to the address from where the request came.
To execute a query via email, the user first retrieves the abstract query form with:
action=http://adsabs.harvard.edu/
abstract_service.html
This will return an executable form. This form can be returned to the ADS in an email message to execute the query. The user enters input for the different fields as required in the forms template. For forms tags like checkboxes or radio buttons, the user can uncomment the appropriate line in the form. Comments in the form that are included with each forms field provide guidance for completing the form before submission. The text part of the form is formatted as comment lines so that the user does not have to modify any irrelevant parts of the form. The retrieval method is already set appropriately.
The recently implemented Z39.50 interface ([Lynch 1997]) conforms to the Library of Congress Z39.50 conventions (http://lcweb.loc.gov/z3950/lcserver.html). This allows any library that uses this protocol to access the ADS through their interface. The databases supported through this interface are listed in Table 2, search fields supported are listed in Table 3, supported relationship attributes are listed in Table 4 and the supported structure attributes are listed in Table 5. Table 6 lists the supported record formats and Table 7 shows the supported record syntax.
| Value | Description |
| AST | Astronomy Database. Contains references from |
| Astronomy related articles | |
| INST | Instrumentation Database. Contains references |
| from articles related to Space Instrumentation | |
| PHY | Physics Database. Contains references from |
| Physics related articles | |
| PRE | Preprint Database. Contains references from the |
| Los Alamos preprint server related to Astronomy | |
| GEN | General Database. Contains references from |
| articles unrelated to the other databases |
| Value | Description |
| 1 | Personal name a |
| 1003 | Author a |
| 4 | Title |
| 5 | Title series b |
| 62 | Abstract |
| 31 | Publication Date |
| 1011 | Entry Date in Database |
| Value | Description | Fields Supporting |
| 3 | Equal a | All |
| 2 | Less than or Equal | Publication Date |
| 4 | Greater than or Equal | Publication Date, Entry |
| Date | ||
| 102 | Relevance b | Title, Abstract, Author |
| Value | Description | Fields Supporting |
| 1 | quoted phrase | Title, Abstract |
| 2 | word | Title, Abstract, Author, Series |
| 6 | word list | Title, Abstract, Author, Series |
| 5 | date | Publication Date, Entry Date |
| Value | Description |
| B | Brief Records (Title, Authors) |
| F | Full Records (all available information) |
| Value | Description |
| 1.2.840.10003.5.101 | SUTRS Records |
| 1.2.840.10003.5.109.3 | HTML Records |
| 1.2.840.10003.5.1000.147.1 | ADS Tagged Records |
Each table notes which search fields support which attribute. The relationship attributes "equal'' and "relevance'' are used to search without and with synonym replacement respectively. If the structure attribute "Phrase'' is selected, the input is considered to be one phrase and is not separated into individual words. If the structure attribute "Word'' is selected and several words are specified, the input is treated as if the structure attribute "Word List'' were specified.
As output, brief and full records are supported. These record formats are the same as in the short results list and in the full abstract display as described above. The supported record syntax is either SUTRS (Simple Unstructured Text Record Syntax), HTML (HyperText Markup Language), or the ADS tagged format. In the HTML record syntax, links to other supported ADS internal and external data sources are included.
A description of this interface is available on-line at:
http://adsabs.harvard.edu/abs_doc/ads_server.html
An example of the ADS Z39.50 interface can be accessed from the Library of Congress Z39.50 Gateway at:
http://lcweb.loc.gov/z3950/
Copyright The European Southern Observatory (ESO)